Flood risk prediction system and method based on multi-source data

Abstract

The invention discloses a flood risk prediction system and method based on multi-source data, and relates to the technical field, and the system comprises a site module generation module, a historical database construction module, a to-be-optimized index value analysis module, an early warning site determination module, an association response model construction module, and an optimal site information output module. The station module generation module is used for generating station modules from all monitoring stations in the same to-be-monitored area according to the corresponding monitoring station information; the historical database construction module is used for constructing a historical database for implementing prediction based on the flood risk prediction system, and the to-be-optimized index value analysis module is used for analyzing and evaluating to-be-optimized index values of various monitoring stations in the station modules of the same to-be-monitored area; the early warning station determination module is used for outputting the type of a monitoring station needing early warning in the to-be-monitored area as an early warning station; and the association response model construction module is used for constructing association response models of various early warning sites and contrast monitoring sites.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of risk prediction, in particular to a flood risk prediction system and method based on multi-source data. BACKGROUND

[0002] Global climate change has led to frequent occurrence of extreme rainfall events, and flood disasters have become one of the main natural disasters threatening human life and property safety and the sustainable development of social economy. Accurate and timely flood monitoring and early warning are crucial to reducing disaster losses, and a scientific and reasonable monitoring site layout is the basis for efficient monitoring. At present, most of the existing flood monitoring sites are laid out based on historical disaster data, topographic features and experience. However, with the acceleration of urbanization, the underlying surface conditions in cities have changed significantly, and the increase in impervious area has intensified the risk of waterlogging. At the same time, factors such as water conservancy construction and land use changes in the basin have also changed the rules of flood evolution. These dynamic changes have gradually exposed many problems in the traditional monitoring site layout mode based on fixed experience. On the one hand, the number of monitoring sites in some high-risk areas (such as urban low-lying areas and river bends) is insufficient, and key flood data cannot be captured in time, resulting in delayed disaster warning. On the other hand, there are phenomena of excessive concentration or unreasonable location of monitoring sites in some areas, causing waste of monitoring resources and difficulty in covering all flood risk points. In addition, the existing monitoring site layout lacks effective combination of real-time data and dynamic risk assessment. The traditional layout method cannot adapt to the monitoring needs under different rainfall conditions and flood conditions, and cannot dynamically optimize and adjust the number and location of sites according to real-time monitoring data and flood risk evolution trends. This makes the reliability and accuracy of the monitoring system greatly discounted when facing sudden extreme flood disasters, and cannot provide accurate and effective data support for flood control decision-making. Therefore, how to scientifically optimize the number and location of the laid flood monitoring sites based on multi-source data and the layout method effectively predicting the event records in the historical database has become a technical problem to be solved to improve the flood monitoring and early warning capability and enhance the regional flood control and disaster reduction level. SUMMARY

[0003] The purpose of the present application is to provide a flood risk prediction system and method based on multi-source data to solve the problems in the prior art.

[0004] To achieve the above purpose, the present application provides the following technical solution: a flood risk prediction method based on multi-source data, the method comprising: Step S100: Extract all monitoring site information laid in the monitored area, the monitoring site information including the function type and location of the monitoring site; generate a site module with the corresponding monitoring site information for all monitoring sites in the same monitored area; Step S200: Construct a historical database based on the flood risk prediction system to implement prediction, and the historical database stores effective prediction events of the flood risk, where the effective prediction event refers to an event in which the deviation between the prediction result of an index of the prediction system and the actual monitoring data is less than the deviation threshold of the corresponding index; Step S300: Extract the monitoring data recorded by the site module in the to-be-monitored region, match the monitoring region in the historical database that meets the similar conditions when the site module is independently analyzed as a control monitoring region, extract the monitoring data recorded by each type of monitoring station in the control monitoring region as control data, and analyze and evaluate the to-be-optimized index values of each type of monitoring station in the site module of the same to-be-monitored region; Step S400: Based on the to-be-optimized index values of each type of monitoring station, comprehensively analyze the abnormal early warning values of the corresponding to-be-monitored region and perform early warning response based on the abnormal early warning values; and output the monitoring station type that needs to be warned in the to-be-monitored region as a warning station; Step S500: Construct an associated response model of each type of warning station and the control monitoring station; filter the optimization information of the warning station based on the associated response model; and traverse all types of warning stations in the same to-be-monitored region to output the corresponding best optimization station information.

[0005] Further, all monitoring stations in the same to-be-monitored region are generated into a site module with corresponding monitoring station information, including the following steps: The same type of monitoring station is taken as the same type of monitoring station, and a rectangular coordinate system is established with the center point of the to-be-monitored region; different types of monitoring stations are distinguished and marked in the rectangular coordinate system, and the same type of monitoring station is taken as a group of stations; the straight-line distance of adjacent monitoring stations in the coordinate system is obtained from any coordinate axis direction of the coordinate system; the adjacent monitoring stations are connected in turn, and the straight-line distance is marked on the connecting line of the adjacent stations to form a site module of the same type of station in the corresponding to-be-monitored region; the site module also records the position coordinates and the number of the same type of monitoring station.

[0006] Further, step S300 includes the following specific process: Step S310: The monitoring data refers to various target data required to be monitored by the corresponding site module; a certain type of site module is determined as a target analysis station, and other types of monitoring stations in the same to-be-monitored region are determined as to-be-analyzed stations; and the similar conditions are as follows: First, all types of monitoring stations in the control monitoring region are the same type of station as the to-be-monitored region, and the number of to-be-analyzed stations is the same as the number of the same type of station in the corresponding control monitoring region; second, the same type of station as the to-be-analyzed station in the control monitoring region respectively satisfies the position similarity of the corresponding site module being greater than a similarity threshold; the position similarity refers to the distance similarity of adjacent stations and the azimuth angle similarity formed in the coordinate system. Both conditions are met to meet similar conditions; Step S320: The to-be-monitored area is divided according to a grid, the number of corresponding target analysis sites in each grid is counted, a flood risk area planning map stored by the system is superimposed, a high-risk area is marked, the number N1 of target analysis sites in the corresponding grid of the high-risk area is obtained, and the total number N0 of target analysis sites in the to-be-monitored area is obtained, and the risk matching degree K1 of the to-be-monitored area corresponding to the target analysis site is calculated by using the formula: K1=(N1 / S1)-(N0 / S0), wherein S1 represents the area in the high-risk area, and S0 represents the total area of the to-be-monitored area; A risk matching degree threshold K0 is set, the risk matching degrees K2 of the same type of monitoring sites of the target analysis site in each control monitoring area are calculated, when K1≤K0, the first characteristic value of the target analysis site is output as 1; when K1>K0, the difference value of |K1-K2| is calculated, the number N2 of control monitoring areas with a difference value greater than a first difference threshold from the target analysis site is extracted, the first characteristic value P1 is calculated, P1=N2 / N3, and N3 represents the number of all control monitoring areas corresponding to the target analysis site; when the first characteristic value is output as 1, it is indicated that the target analysis site arranged in the high-risk area does not meet the requirements and the density is low, and optimization is required; when the threshold K0 is greater than the threshold K0, the relationship with the control monitoring area is further analyzed, in order to quantify the deviation relationship of the target analysis site of this type of monitoring site in different distribution states on the risk matching degree of the monitoring data; Step S330: The event in which the data peak of the target analysis site recorded in the historical database in the flood occurrence event is an effective monitoring event, the total number M1 of flood occurrence events monitored by the target analysis site in the historical database is extracted, the extreme event capture rate Z1 of the target analysis site is calculated, Z1=M0 / M1, wherein M0 represents the number of effective monitoring events; An extreme event capture rate threshold Z0 is set, the extreme event capture rates Z2 of the same type of monitoring sites of the target analysis site in each control monitoring area are calculated, when Z1≤Z0, the second characteristic value of the target analysis site is output as 1; when Z1>Z0, the difference value of |Z1-Z2| is calculated, the number M2 of control monitoring areas with a difference value greater than a second difference threshold from the target analysis site is extracted, the second characteristic value P2 is calculated, P2=M2 / N3; Step S340: Based on the first characteristic value P1 and the second characteristic value P2, the to-be-optimized index value Q of the corresponding type of target analysis site in each site module of the to-be-monitored area is calculated, Q=a1*P1+a2*P2; wherein a1 and a2 represent corresponding reference coefficients.

[0007] The greater the to-be-optimized index is, the greater the possibility that the target analysis site has an unreasonable layout is; and the analysis of each type of monitoring site is realized based on control variables in the historical database, and in the calculation of the to-be-optimized index, the rationality of the layout of the same type of monitoring site in the monitoring area is evaluated, while in the foregoing direct judgment of the characteristic value, the layout of the output site is unreasonable.

[0008] Further, step S400 comprises the following: A to-be-optimized index value threshold Q0 is set, the to-be-optimized index values of each type of monitoring site in the same to-be-monitored area are compared with the size of the to-be-optimized index value threshold, and the type of monitoring site corresponding to Q≥Q0 is marked as an abnormal site; an abnormal early warning value U of the to-be-monitored area is calculated, U=m1 / m2; m1 represents the number of types of abnormal sites, and m2 represents the total number of types of monitoring sites in the to-be-monitored area; When U=0, no early warning is performed; when U≠0, the corresponding type of abnormal site is early warned; The monitoring site that needs to be early warned is the abnormal site that is marked.

[0009] Further, step S500 comprises the following: Step S510: the type number N2 of the control monitoring site recorded when the early warning site analyzes the first characteristic and the corresponding risk matching degree difference g1=|K1-K2| are obtained, the minimum area d1 of all sites of the same type corresponding to the early warning site is marked, and the site density f1 of the area where the early warning site is located is calculated, f1=h1 / d1, wherein h1 represents the total number of all sites of the same type corresponding to the early warning site; the site density f2 of the site of the area formed by each control monitoring site and the early warning site of the same type is calculated; the density difference z1 is obtained, z1=|f1-f2|; The first data set B1 formed by the early warning site and each control monitoring site is sequentially corresponding, B1=(z1, g1), the density difference of the early warning site and each control monitoring site corresponds to the risk matching degree difference to form a data set, which can effectively and clearly quantify the monitoring analysis from the data impact level under different layout information, so as to evaluate whether the layout of the early warning site is reasonable and the optimization direction; the correlation response model r1 of the early warning site and N2 types of control monitoring sites is calculated, r1=∑[(z 1i- z 10 ) g1i -g 10 )] / [∑(z 1i -z 10 ) 2 ∑(g 1i -g 10 ) 2 ] 1 / 2 ; wherein z 1i , g1i represents the i-th density difference value and the risk matching degree difference value in the first data set B, z 10 , g 10 represents the average value of the corresponding density difference value and the average value of the risk matching degree difference value in all data sets; Step S520: Similarly, calculate the correlation response model r2 of the early warning site when analyzing the second feature and the corresponding M2 type control monitoring site; compare r1 and r2, and select the control monitoring site corresponding to the analysis with the maximum correlation response model output value and greater than the correlation coefficient threshold value as the monitoring site under investigation; the layout information corresponding to the monitoring site under investigation is used as the optimization information; The purpose of analyzing the correlation coefficient that is the largest and greater than the threshold value is to effectively screen out which control monitoring sites have a layout density influence evaluation index value size relationship with the early warning site; the larger the correlation coefficient, the greater the influence of the layout information on the index evaluation; the stronger the directionality and accuracy that can be optimized; Step S530: Mark the monitoring area where each monitoring site under investigation is located as an investigation area; each type of early warning site corresponds to a set of investigation areas; traverse all early warning sites to generate a corresponding set of investigation areas; if there is an intersection between each set of investigation areas, output the optimization information of the investigation area record corresponding to the intersection as the best optimization site information; If there is no intersection, for each type of early warning site corresponding to the investigation area, select the optimization information corresponding to the least number of variable sites as the best optimization site information; if the correlation response model output value is less than or equal to the correlation coefficient threshold value, only the early warning response of the early warning site type is performed.

[0010] A flood risk prediction system based on multi-source data, the system comprising a site module generation module, a historical database construction module, a to-be-optimized index value analysis module, an early warning site determination module, a correlation response model construction module, and a best optimization site information output module; The site module generation module is used to generate a site module with corresponding monitoring site information for all monitoring sites in the same to-be-monitored area; The historical database construction module is used to construct a historical database based on the flood risk prediction system for implementation of prediction, The to-be-optimized index value analysis module is used to analyze and evaluate the to-be-optimized index values of various types of monitoring sites in the site module of the same to-be-monitored area; The early warning site determination module is used to comprehensively analyze the abnormal early warning values of the corresponding to-be-monitored area and perform an early warning response based on the abnormal early warning values; the type of monitoring site that needs to be early warned in the to-be-monitored area is output as an early warning site; The correlation response model construction module is used to construct a correlation response model of various types of early warning sites and control monitoring sites; The optimal optimization site information output module is used for traversing all types of early warning sites in the same to-be-monitored region and outputting corresponding optimal optimization site information.

[0011] Further, the to-be-optimized index value analysis module comprises a site type determination unit, a first characteristic value calculation unit, a second characteristic value calculation unit and a to-be-optimized index value calculation unit. The site type determination unit is used for determining that a site module of a certain type is a target analysis site and other types of monitoring sites in the same to-be-monitored region are to-be-analyzed sites. The first characteristic value calculation unit is used for calculating a first characteristic value of the target analysis site and a control monitoring region based on a risk matching degree. The second characteristic value calculation unit is used for calculating a second characteristic value of the target analysis site and the control monitoring region based on an extreme event capture rate. The to-be-optimized index value calculation unit is used for calculating a to-be-optimized index value based on the first characteristic value and the second characteristic value.

[0012] Further, the correlation response model construction module comprises a density difference calculation unit, a characteristic value difference calculation unit and a correlation response model output comparison unit. The density difference calculation unit calculates site densities of region sites formed by sites of the same type of each control monitoring site and the early warning site and calculates a density difference. The characteristic value difference calculation unit is used for calculating a difference based on a risk matching degree and calculating a difference based on an extreme event capture rate. The correlation response model output comparison unit is used for calculating a correlation coefficient based on a data group and selecting a control monitoring site corresponding to analysis as an investigation monitoring site when a correlation response model output value is maximum and greater than a correlation coefficient threshold value; layout information corresponding to the investigation monitoring site is used as optimization information; and optimal optimization site information is output by comparison and analysis.

[0013] Compared with the prior art, the present application has the following beneficial effects: The present application can accurately identify the insufficiency of site layout in quantity and position by extracting site information of a to-be-monitored region to generate a site module, matching a control monitoring region in combination with effective prediction events in a historical database and analyzing a to-be-optimized index value. The historical database constructed by the present application stores effective prediction events, can dynamically evaluate the rationality of site layout based on multi-source data by comparing and analyzing to-be-monitored region data and control region data, and can match a control region again, calculate a to-be-optimized index and further determine an optimization scheme when the environment of the to-be-monitored region changes, so that the site layout can adapt to the change of the flood evolution law of a basin and the problem that the traditional layout cannot be dynamically adjusted is solved, thereby enhancing the adaptability of the monitoring system to environmental changes.

[0014] By evaluating the values ​​of the indicators to be optimized, this method can identify sites with inappropriate layouts. For high-risk areas with insufficient site density, it recommends adding new sites. For redundant sites with low contribution, it recommends migrating or replacing them, thereby optimizing resource allocation while ensuring effective monitoring. For example, by calculating the correlation coefficient between the difference in site density and the difference in risk matching, the optimal optimization information can be screened, achieving the best monitoring effect with the minimum number of sites, avoiding resource waste, and improving the overall efficiency of the monitoring system. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a flow chart of a flood risk prediction method based on multi-source data according to the present invention. DETAILED DESCRIPTION

[0016] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0017] Example: Figure 1 As shown, the present invention provides a flood risk prediction method based on multi-source data, the method comprising: Step S100: extracting information of all monitoring sites deployed in the area to be monitored, the monitoring site information including the function type and location of the monitoring site; generating a site module for all monitoring sites in the same area to be monitored with the corresponding monitoring site information; Step S200: Constructing a historical database based on the flood risk prediction system. The historical database stores effective flood risk prediction events. An effective prediction event refers to an event in which the deviation between the indicator prediction result of the prediction system and the actual monitoring data is less than the corresponding indicator deviation threshold. Step S300: extracting monitoring data recorded by the site modules in the monitored area, matching monitoring areas in the historical database that meet similar conditions as the site modules when analyzed independently as control monitoring areas, extracting monitoring data recorded by various monitoring sites in the control monitoring area as control data, and analyzing and evaluating the values ​​of indicators to be optimized for various monitoring sites in the site modules in the same monitored area; Step S400: Based on the values ​​of the indicators to be optimized of various monitoring sites, comprehensively analyze the abnormal warning values ​​of the corresponding area to be monitored and perform an early warning response based on the abnormal warning values; output the type of monitoring site requiring early warning in the area to be monitored as an early warning site; Step S500: constructing an association response model between various types of warning sites and control monitoring sites; filtering optimization information of warning sites based on the association response model; traversing all types of warning sites in the same monitored area and outputting corresponding best optimized site information.

[0018] All monitoring sites in the same to-be-monitored region are generated into a site module with corresponding monitoring site information, including the following steps: The monitoring sites of the same function type are taken as the same type of monitoring sites, and a rectangular coordinate system is established with the center point of the to-be-monitored region; different types of monitoring sites are distinguished and marked in the rectangular coordinate system, and the same type of monitoring sites is taken as a group of sites, and the straight-line distance of adjacent monitoring sites on the coordinate system is obtained from any coordinate axis direction of the coordinate system; the adjacent monitoring sites are connected in turn, and the straight-line distance is marked on the connecting line of the adjacent sites to form a site module of the same type in the corresponding to-be-monitored region; the site module also records the position coordinates and the number of the same type of monitoring sites.

[0019] The adjacent monitoring sites from any coordinate axis direction of the coordinate system means that only the coordinate data of the monitoring sites corresponding to one coordinate axis is observed, for example, from the negative direction of the x-axis, the coordinates of the same type of monitoring sites are recorded as a (-2, 3), b (1, 5) and c (3, -1); then the connection order in turn is abc.

[0020] Step S300 includes the following specific process: Step S310: The monitoring data refers to various target data required to be monitored by the corresponding site module; a certain type of site module is determined as a target analysis site, and other types of monitoring sites in the same to-be-monitored region are determined as to-be-analyzed sites; the similar conditions are met as follows: First, the same type of monitoring sites in the to-be-monitored region is compared with the to-be-analyzed sites, and the number of the to-be-analyzed sites is the same as the number of the same type of sites in the corresponding monitoring region; second, the same type of sites in the to-be-analyzed sites in the monitoring region should meet the position similarity of the corresponding site module greater than the similarity threshold; the position similarity refers to the distance similarity of adjacent sites and the azimuth angle similarity formed in the coordinate system; Both of the above conditions are met to meet the similar conditions; Step S320: The to-be-monitored region is divided according to the grid, the number of corresponding target analysis sites in each grid is counted, the flood risk area planning map stored in the system is superimposed, the high-risk area is marked, the number N1 of target analysis sites in the corresponding grid of the high-risk area is obtained, and the total number N0 of target analysis sites in the to-be-monitored region is obtained, and the formula K1=(N1 / S1)-(N0 / S0) is used to calculate the risk matching degree K1 of the corresponding target analysis sites in the to-be-monitored region, wherein S1 represents the area in the high-risk area, and S0 represents the total area of the to-be-monitored region; A risk matching degree threshold K0 is set, the risk matching degree K2 of each control monitoring area and the same type monitoring station of the target analysis station is calculated, when K1≤K0, the first characteristic value of the target analysis station is output as 1; when K1>K0, the difference value of |K1-K2| is calculated, the number N2 of the control monitoring area with the difference value greater than the first difference threshold from the target analysis station is extracted, the first characteristic value P1 is calculated, P1=N2 / N3, N3 represents the number of all control monitoring areas corresponding to the target analysis station; when the first characteristic value is output as 1, it indicates that the target analysis station arranged in the high-risk area does not meet the requirements, the density is low, and optimization is needed; when greater than the threshold K0, the relationship with the control monitoring area is further analyzed in order to quantify the deviation relationship of the target analysis station in the type of monitoring station on the risk matching degree of the monitoring data in different distribution states; Step S330: The event in which the data peak in the flood occurrence event recorded in the historical database of the target analysis station is an effective monitoring event, the total number M1 of flood occurrence events monitored by the target analysis station in the historical database is extracted, the extreme event capture rate Z1 of the target analysis station is calculated, Z1=M0 / M1, wherein M0 represents the number of effective monitoring events; the definition of effective monitoring event is that according to actual requirements, if any one of the actual same type monitoring station captures the data peak in the flood occurrence event, the monitoring station can be regarded as effective data, then this event is an effective monitoring event, if the actual same type monitoring station needs multiple monitoring stations to capture the data peak to be effective, and the history only records that one station captures, then this event is not regarded as an effective monitoring event; the target analysis station is a group of stations and is not independent; An extreme event capture rate threshold Z0 is set, the extreme event capture rate Z2 of each control monitoring area and the same type monitoring station of the target analysis station is calculated, when Z1≤Z0, the second characteristic value of the target analysis station is output as 1; when Z1>Z0, the difference value of |Z1-Z2| is calculated, the number M2 of the control monitoring area with the difference value greater than the second difference threshold from the target analysis station is extracted, the second characteristic value P2 is calculated, P2=M2 / N3; Step S340: Based on the first characteristic value P1 and the second characteristic value P2, the optimization index value Q of the corresponding type target analysis station in each station module of the to-be-monitored area is calculated, Q=a1*P1+a2*P2; wherein a1 and a2 represent the corresponding reference coefficients.

[0021] The larger the optimization index is, the greater the possibility of unreasonable layout of the target analysis station is; and the control variables in the historical database are used to realize the targeted analysis of each type of monitoring station, the rationality of the layout of the same type monitoring station of the control monitoring area is evaluated in the calculation of the optimization index, and the layout of the output station is unreasonable in the foregoing direct judgment of the characteristic value.

[0022] Step S400 includes the following: A threshold value Q0 of the to-be-optimized index value is set, and the threshold value of the to-be-optimized index value corresponding to different monitoring sites is the same. Generally, Q0 = 1 is set. The to-be-optimized index value of each type of monitoring site in the same to-be-monitored region is compared with the to-be-optimized index value threshold, and the type of monitoring site corresponding to Q ≥ Q0 is marked as an abnormal site. An abnormal early warning value U of the to-be-monitored region is calculated, U = m1 / m2; m1 represents the number of types of abnormal sites, and m2 represents the total number of types of monitoring sites in the to-be-monitored region. When U = 0, no early warning is performed; when U ≠ 0, an early warning response is performed on the corresponding type of abnormal site. The monitoring site that needs to be warned is the marked abnormal site.

[0023] Step S500 includes the following: Step S510: The type number N2 of the contrast monitoring site recorded by the early warning site when analyzing the first feature and the corresponding risk matching degree difference g1 = |K1-K2| are obtained, the minimum area d1 of the region formed by all sites of the same type corresponding to the early warning site is marked, and the site density f1 of the region where the early warning site is located is calculated, f1 = h1 / d1, where h1 represents the total number of all sites of the same type corresponding to the early warning site. The site density f2 of the region formed by the sites of the same type as the early warning site of each contrast monitoring site is calculated. The density difference z1 is obtained, z1 = |f1-f2|. The first data set B1 formed by the early warning site and each contrast monitoring site is sequentially corresponded, B1 = (z1, g1), and the density difference of the early warning site and each contrast monitoring site and the risk matching degree difference form a data set, which can effectively and clearly quantify the monitoring analysis from the data impact level under different layout information, so as to evaluate whether the layout of the early warning site is reasonable and the optimization direction. The correlation response model r1 of the early warning site and the N2 types of contrast monitoring sites is calculated, r1 = ∑[(z 1i- z 10 ) g1i -g 10 )] / [∑(z 1i -z 10 ) 2 ∑(g 1i -g 10 ) 2 ] 1 / 2 ; wherein z 1i , g 1i represent the i-th density difference and risk matching degree difference in the formed first data set B, and z 10 , g 10 represent the average value of the corresponding density difference and the average value of the risk matching degree difference in all data sets. Step S520: Similarly, calculate the correlation response model r2 of the early warning site when analyzing the second feature and the corresponding M2 type control monitoring site; compare r1 and r2, select the control monitoring site corresponding to the analysis of the correlation response model output value maximum and greater than the correlation coefficient threshold as the monitoring site under investigation; the layout information corresponding to the monitoring site under investigation is used as the optimization information; The purpose of analyzing the correlation coefficient maximum and greater than the threshold value is to effectively screen out which control monitoring sites have layout density influence evaluation index value size relationship with the early warning site; the greater the correlation coefficient, the greater the influence of the layout information on the index evaluation; the stronger the directionality and accuracy of optimization; As shown in the embodiment: obtain the number M2 of control monitoring sites recorded by the early warning site when analyzing the second feature and the corresponding extreme event capture rate difference g2=|Z1-Z2|, calculate the site density f3 of each control monitoring site and the same type site of the early warning site; obtain the density difference z2, z2=|f1-f3|; the second data set B2 formed by the early warning site and each control monitoring site is corresponded in turn, B2=(z2, g2), and the correlation coefficient r2 of the early warning site and the M2 control monitoring sites is calculated; Step S530: Mark the monitoring area where each monitoring site under investigation is located as an investigation area, and each type of early warning site corresponds to a set of investigation areas; traverse all early warning sites to generate corresponding sets of investigation areas; if there is an intersection between each set of investigation areas, output the optimization information recorded by the intersection corresponding investigation area as the best optimization site information; the optimization information refers to the position layout relationship of each type of monitoring site recorded by the area, such as the information recorded by the site module analyzed above.

[0024] If there is no intersection, select the optimization information corresponding to the least number of variable sites for each type of early warning site corresponding investigation area as the best optimization site information of itself; if the correlation response model output value is less than or equal to the correlation coefficient threshold, only the early warning response of the early warning site type is performed.

[0025] A flood risk prediction system based on multi-source data, the system comprising a site module generation module, a historical database construction module, a to-be-optimized index value analysis module, an early warning site determination module, a correlation response model construction module, and a best optimization site information output module; The site module generation module is used to generate a site module by using corresponding monitoring site information for all monitoring sites in the same to-be-monitored area; The historical database construction module is used to construct a historical database based on the flood risk prediction system for implementation of prediction, The to-be-optimized index value analysis module is used to analyze and evaluate the to-be-optimized index values of each type of monitoring site in the site module of the same to-be-monitored area; The warning site determination module is used to comprehensively analyze the abnormal warning values ​​of the corresponding monitored area and make warning responses based on the abnormal warning values; output the type of monitoring site that needs warning in the monitored area as a warning site; The association response model construction module is used to construct the association response model of various early warning sites and control monitoring sites; The best optimization site information output module is used to traverse all types of warning sites in the same monitored area and output the corresponding best optimization site information.

[0026] The index value analysis module to be optimized includes a site type determination unit, a first characteristic value calculation unit, a second characteristic value calculation unit and an index value calculation unit to be optimized; The site type determination unit is used to determine a certain type of site module as a target analysis site, and other types of monitoring sites in the same monitored area as sites to be analyzed; The first eigenvalue calculation unit is used to calculate the first eigenvalue of the target analysis site and the control monitoring area based on the risk matching degree; The second eigenvalue calculation unit is used to calculate the second eigenvalue of the target analysis site and the control monitoring area based on the extreme event capture rate; The to-be-optimized index value calculation unit is configured to calculate the to-be-optimized index value based on the first eigenvalue and the second eigenvalue.

[0027] The associated response model construction module includes a density difference calculation unit, a characteristic value difference calculation unit and an associated response model output comparison unit; The density difference calculation unit calculates the site density of regional sites of the same type as the control monitoring sites and the early warning sites, and calculates the density difference; The characteristic value difference calculation unit is used to calculate the difference based on the risk matching degree and the extreme event capture rate; The correlation response model output comparison unit is used to construct a data group to calculate the correlation coefficient; and select the control monitoring site corresponding to the analysis when the correlation response model output value is the largest and greater than the correlation coefficient threshold as the inspection monitoring site; the layout information corresponding to the inspection monitoring site is used as the optimization information; and the comparison analysis outputs the best optimized site information.

[0028] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art will be able to modify the technical solutions described in the aforementioned embodiments or substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A flood risk prediction method based on multi-source data, characterized by: The method comprises: Step S100: extracting information of all monitoring sites deployed in the area to be monitored, wherein the information includes the function type and location of the monitoring site; generating a site module for all monitoring sites in the same area to be monitored with the corresponding monitoring site information; Step S200: constructing a historical database for implementing predictions based on the flood risk prediction system, wherein the historical database stores effective flood risk prediction events, wherein the effective prediction events refer to events in which the deviations between the indicator prediction results of the prediction system and the actual monitoring data are less than the corresponding indicator deviation threshold; Step S300: extracting monitoring data recorded by the site modules in the monitored area, matching monitoring areas in the historical database that meet similar conditions as the site modules when analyzed independently as control monitoring areas, extracting monitoring data recorded by various monitoring sites in the control monitoring area as control data, and analyzing and evaluating the values ​​of indicators to be optimized for various monitoring sites in the site modules in the same monitored area; Step S400: Based on the values ​​of the indicators to be optimized of various monitoring sites, comprehensively analyze the abnormal warning values ​​of the corresponding area to be monitored and perform an early warning response based on the abnormal warning values; output the type of monitoring site requiring early warning in the area to be monitored as an early warning site; Step S500: constructing an association response model between various types of warning sites and control monitoring sites; filtering optimization information of warning sites based on the association response model; traversing all types of warning sites in the same monitored area and outputting corresponding best optimized site information.

2. The flood risk prediction method based on multi-source data according to claim 1, characterized in that: The step of generating a site module using corresponding monitoring site information for all monitoring sites in the same monitored area includes the following steps: Monitoring sites with the same functional type are regarded as the same type of monitoring sites, and a rectangular coordinate system is established with the center point of the area to be monitored; different types of monitoring sites are distinguished and marked in the rectangular coordinate system, and the same type of monitoring sites are grouped as a group of sites, and the straight-line distance between adjacent monitoring sites in the coordinate system is obtained from any coordinate axis direction of the coordinate system; adjacent monitoring sites are connected in sequence, and the straight-line distance is marked on the connecting line of adjacent sites to form a site module of the same type corresponding to the area to be monitored; the site module also records the location coordinates and number of monitoring sites of the same type.

3. The flood risk prediction method based on multi-source data according to claim 1, characterized in that: The step S300 includes the following specific processes: Step S310: The monitoring data refers to various target data that the corresponding site module needs to monitor; a certain type of site module is determined as the target analysis site, and other types of monitoring sites in the same monitored area are sites to be analyzed; the similarity conditions are as follows: First, all types of monitoring sites in the control monitoring area must be of the same type as those in the monitored area, and the number of sites to be analyzed must be the same as the number of sites of the same type in the corresponding control monitoring area. Second, sites of the same type as those to be analyzed in the control monitoring area must satisfy the positional similarity of their respective corresponding site modules greater than a similarity threshold. Positional similarity refers to the similarity formed by the distance similarity between adjacent sites and the similarity of the azimuth angles in the coordinate system. The above two conditions are both met to meet the similarity condition; Step S320: Divide the area to be monitored into grids, count the number of target analysis sites corresponding to each grid, overlay the flood risk area planning map stored in the system, mark the high-risk areas, obtain the number N1 of target analysis sites in the grids corresponding to the high-risk areas, and the total number N0 of target analysis sites in the area to be monitored, and calculate the risk matching degree K1 of the target analysis sites corresponding to the area to be monitored using the formula: K1=(N1 / S1)-(N0 / S0), where S1 represents the area within the high-risk area and S0 represents the total area of ​​the area to be monitored; Set the risk matching threshold K0, calculate the risk matching degree K2 of each control monitoring area and the same type of monitoring site as the target analysis site, when K1≤K0, output the first eigenvalue of the target analysis site as 1; when K1>K0, calculate the difference of |K1-K2|, extract the number N2 of control monitoring areas whose difference with the target analysis site is greater than the first difference threshold, and calculate the first eigenvalue P1, P1=N2 / N3, N3 represents the number of all control monitoring areas corresponding to the target analysis site; Step S330: Counting the flood events with data peaks recorded by the target analysis site in the historical database as valid monitoring events, extracting the total number of flood events M1 monitored by the target analysis site in the historical database, and calculating the extreme event capture rate Z1 of the target analysis site, where Z1 = M0 / M1, where M0 represents the number of valid monitoring events. Set the extreme event capture rate threshold Z0, calculate the extreme event capture rate Z2 of each control monitoring area and the same type of monitoring site as the target analysis site, and when Z1≤Z0, output the second eigenvalue of the target analysis site as 1; when Z1>Z0, calculate the difference of |Z1-Z2|, extract the number M2 of control monitoring areas whose difference with the target analysis site is greater than the second difference threshold, and calculate the second eigenvalue P2, P2=M2 / N3; Step S340: Based on the first eigenvalue P1 and the second eigenvalue P2, calculate the index value Q to be optimized for the corresponding type of target analysis site in each site module in the monitored area, Q=a1*P1+a2*P2; where a1 and a2 represent corresponding reference coefficients.

4. The flood risk prediction method based on multi-source data according to claim 3, characterized in that: The step S400 includes the following: Set the threshold Q0 of the indicator value to be optimized, compare the values ​​of the indicator value to be optimized of each type of monitoring station in the same monitoring area with the threshold value of the indicator value to be optimized, and mark the monitoring station with the corresponding type of Q≥Q0 as an abnormal station; calculate the abnormal warning value U of the monitoring area to be monitored, U=m1 / m2; m1 represents the number of types of abnormal stations, and m2 represents the total number of types of monitoring stations in the monitoring area to be monitored; When U=0, no warning is given; When U≠0, an early warning response is issued to the corresponding type of abnormal site; The monitoring sites requiring early warning are marked abnormal sites.

5. The flood risk prediction method based on multi-source data according to claim 3 is characterized by: The step S500 includes the following: Step S510: Obtain the number of control monitoring site types N2 and the corresponding risk matching difference g1=|K1-K2| recorded by the warning site when analyzing the first feature, mark the minimum area d1 of all sites of the same type corresponding to the warning site, and calculate the site density f1 of the area where the warning site is located, f1=h1 / d1, where h1 represents the total number of all sites of the same type corresponding to the warning site; calculate the site density f2 of the regional sites of the same type as each control monitoring site and the warning site; and obtain the density difference z1, z1=|f1-f2|; The first data group B1 consisting of the early warning site and each control monitoring site is matched in sequence, B1=(z1,g1), and the correlation response model r1 of the early warning site and the N2 type control monitoring site is calculated. r1=∑[(z 1i- z 10 )( g1i -g 10 )] / [∑(z 1i -z 10 ) 2 ∑(g 1i -g 10 ) 2 ] 1 / 2 ; where z 1i 、g 1i Indicates the density difference and risk matching difference of the first data group B, z 10 、g 10 It represents the average value of the corresponding density difference and the average value of the risk matching difference in all data groups; Step S520: Similarly, calculate the correlation response model r2 between the warning site and the corresponding M2 control monitoring site when analyzing the second feature; compare r1 and r2, and select the corresponding control monitoring site when the correlation response model output value is the largest and greater than the correlation coefficient threshold as the inspection monitoring site; and use the layout information corresponding to the inspection monitoring site as optimization information; Step S530: Mark the monitoring area where each inspection monitoring site is located as an inspection area, and each type of early warning site corresponds to a set of inspection areas; Traverse all warning sites and generate corresponding inspection area sets respectively; if there is an intersection among the inspection area sets, the optimization information recorded in the inspection area corresponding to the intersection is output as the best optimized site information; If there is no intersection, the optimization information corresponding to the inspection area with the least number of changed sites is selected as the best optimized site information for each type of warning site. If the output values ​​of the correlation response model are all less than or equal to the correlation coefficient threshold, only the warning site type will be responded to.

6. A flood risk prediction system based on multi-source data, comprising: The system includes a site module generation module, a historical database construction module, an optimization index value analysis module, an early warning site determination module, an associated response model construction module and an optimal optimization site information output module; The site module generation module is used to generate site modules for all monitoring sites in the same monitored area with corresponding monitoring site information; The historical database construction module is used to construct a historical database based on the flood risk prediction system. The to-be-optimized index value analysis module is used to analyze and evaluate the to-be-optimized index values ​​of various monitoring sites in the site modules of the same to-be-monitored area; The warning site determination module is used to comprehensively analyze the abnormal warning values ​​of the corresponding monitored area and perform warning responses based on the abnormal warning values; Output the monitoring site type that needs early warning in the monitoring area as early warning site; The association response model construction module is used to construct association response models of various early warning sites and control monitoring sites; The best optimized site information output module is used to traverse all types of warning sites in the same monitored area and output corresponding best optimized site information.

7. The flood risk prediction system based on multi-source data according to claim 6, characterized in that: The to-be-optimized index value analysis module includes a site type determination unit, a first characteristic value calculation unit, a second characteristic value calculation unit, and a to-be-optimized index value calculation unit; The site type determination unit is used to determine a certain type of site module as a target analysis site, and other types of monitoring sites in the same monitored area as sites to be analyzed; The first characteristic value calculation unit is used to calculate the first characteristic value of the target analysis site and the control monitoring area based on the risk matching degree; The second eigenvalue calculation unit is used to calculate the second eigenvalue of the target analysis site and the control monitoring area based on the extreme event capture rate; The to-be-optimized index value calculation unit is configured to calculate the to-be-optimized index value based on the first eigenvalue and the second eigenvalue.

8. The flood risk prediction system based on multi-source data according to claim 7, characterized in that: The associated response model construction module includes a density difference calculation unit, a characteristic value difference calculation unit and an associated response model output comparison unit; The density difference calculation unit calculates the site density of regional sites of the same type as the control monitoring sites and the early warning sites, and calculates the density difference; The characteristic value difference calculation unit is used to calculate the difference based on the risk matching degree and the extreme event capture rate; The correlation response model output comparison unit is used to construct a data set to calculate the correlation coefficient; The control monitoring site corresponding to the analysis when the output value of the correlation response model is the largest and greater than the correlation coefficient threshold is selected as the inspection monitoring site; the layout information corresponding to the inspection monitoring site is used as the optimization information; Comparative analysis outputs the best optimized site information.

Images