A flood risk prediction method, device, equipment and storage medium
By acquiring and classifying precipitation data and water level data, and training the flood prediction model, the problem of low accuracy of flood risk prediction in the existing technology is solved, and more accurate flood risk prediction and early warning capabilities are achieved.
Patent Information
- Application Number
- CN202410456094.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-15
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-04-15
AI Technical Summary
The lack of accurate definition and classification of prior art in flood risk predictions has led to low prediction accuracy, especially in long-term forecast periods and in complex urban drainage systems.
By obtaining the precipitation data of the target area and the water level data of the urban drainage system, determining the event type, and training the flood prediction model based on the data corresponding to multiple target type events in the preset time period, inputting the data of the current time period for prediction, improving The accuracy of flood risk prediction.
Improve the accuracy of flood risk prediction, especially in long-term foreseeable periods and complex urban drainage systems, enhance the ability to identify and early warning of flood risk, reduce the root mean square error and improve the Nash efficiency coefficient.
Smart Images

Figure CN118153455B_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of data processing, and in particular to a flood risk prediction method, device, equipment and storage medium. Background Art
[0002] With the intensification of climate change, urban flood disasters have become more frequent and severe, causing greater adverse impacts on the economic and social development of cities. Therefore, the prediction of urban flood risks is particularly important.
[0003] However, due to the lack of accurate definition and classification of urban flood time, the accuracy of flood risk prediction is low. Therefore, how to improve the accuracy of flood risk prediction has become a technical problem that needs to be solved urgently. Summary of the invention
[0004] In order to solve the above technical problems, the embodiments of the present disclosure provide a flood risk prediction method, device, equipment and storage medium.
[0005] In a first aspect, the present disclosure provides a flood risk prediction method, the method comprising:
[0006] Obtain precipitation data of the target area in the current time period and water level data of the urban drainage system;
[0007] Determine the event type to which the precipitation data in the current time period belongs based on the precipitation data in the current time period and the water level data of the urban drainage system;
[0008] If the event type to which the precipitation data in the current time period belongs is a target type event, the precipitation data in the current time period is input into a flood prediction model corresponding to the target type event, and after being processed by the flood prediction model corresponding to the target type event, the water level data of the urban drainage system of the target area in the prediction time period is output; wherein the flood prediction model corresponding to the target type event is trained based on precipitation data and water level data of the urban drainage system corresponding to a plurality of target type events in the target area within a preset time period, the target type event is used to characterize an event in which the water level in the urban drainage system rises due to precipitation, and the water level change in the urban drainage system exceeds a first preset threshold, and the water level data of the urban drainage system of the target area in the prediction time period is used to characterize the risk of flooding in the target area in the prediction time period.
[0009] In an optional embodiment, the time period corresponding to the target type event includes the time point when the water level rises to exceed the flood level. The time period corresponding to the target type event starts at the time point when the water level rises in the urban drainage system caused by precipitation and the water level change in the urban drainage system first exceeds the first preset threshold, and ends at the time point when the water level in the urban drainage system recedes to the flood level after the precipitation ends.
[0010] In an optional implementation manner, the step of inputting the precipitation data in the current time period into the flood prediction model corresponding to the target type event, and after being processed by the flood prediction model corresponding to the target type event, before outputting the water level data of the urban drainage system of the target area in the prediction time period, further includes:
[0011] Obtaining precipitation data of the target area within a preset time period and water level data of the urban drainage system;
[0012] Classifying the precipitation data of the target area within the preset time period and the water level data of the urban drainage system, and determining the precipitation data and the water level data of the urban drainage system corresponding to a plurality of target type events of the target area within the preset time period;
[0013] Model training is performed based on precipitation data corresponding to a plurality of target type events in the target area within the preset time period and water level data of the urban drainage system to obtain a flood prediction model corresponding to the target type event.
[0014] In an optional implementation manner, before classifying the precipitation data of the target area within the preset time period and the water level data of the urban drainage system, and determining the precipitation data and the water level data of the urban drainage system corresponding to the multiple target type events of the target area within the preset time period, the method further includes:
[0015] If there are missing data in the precipitation data of the target area within the preset time period and / or the water level data of the urban drainage system, interpolation processing is performed on the missing data using a preset interpolation function to obtain interpolated data;
[0016] Based on the interpolation data and the precipitation data of the target area within the preset time period and the water level data of the urban drainage system, the corrected data is determined and used as the precipitation data of the target area within the preset time period and the water level data of the urban drainage system.
[0017] In an optional implementation manner, after classifying the precipitation data of the target area within the preset time period and the water level data of the urban drainage system, and determining the precipitation data and the water level data of the urban drainage system corresponding to a plurality of target type events of the target area within the preset time period, the method further includes:
[0018] Sorting the precipitation data and water level data corresponding to the multiple target type events in the target area within the preset time period according to time information to obtain a target type event sequence;
[0019] Performing interpolation and smoothing processing on each target type event in the target type event sequence to obtain an updated target type event sequence;
[0020] Accordingly, the model training is performed based on the precipitation data and the water level data of the urban drainage system corresponding to the multiple target type events in the target area within the preset time period to obtain the flood prediction model corresponding to the target type event, including:
[0021] The updated target type event sequence is used to perform model training to obtain a flood prediction model corresponding to the target type event.
[0022] In an optional implementation manner, before performing interpolation and smoothing processing on each target type event in the target type event sequence to obtain an updated target type event sequence, the method further includes:
[0023] Performing a correlation analysis on the target type event sequence to determine a maximum mutual correlation coefficient corresponding to the target type event sequence; wherein the maximum mutual correlation coefficient is used to characterize the degree of correlation between the precipitation data and the water level data under a preset lag time;
[0024] Accordingly, performing interpolation and smoothing processing on each target type event in the target type event sequence to obtain an updated target type event sequence includes:
[0025] Before each target type event in the target type event sequence, a first type event under a corresponding preset interception time length is added to obtain a first target type event sequence; wherein the first type event is used to characterize an event in which there is no precipitation and the water level change in the urban drainage system does not exceed a first preset threshold value;
[0026] After each target type event in the first target type event sequence, a second type event with a corresponding preset extension time length is added to obtain a second target type event sequence; wherein the second type event is used to characterize an event in which there is no precipitation and the water level in the urban drainage system drops from the flood level to the base flow level.
[0027] performing a correlation analysis on the second target type event sequence to determine a maximum mutual correlation coefficient corresponding to the second target type event sequence;
[0028] If the difference between the maximum mutual correlation coefficient corresponding to the second target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is less than a preset difference, the second target type event sequence is used as an updated target type event sequence.
[0029] In an optional implementation manner, after performing correlation analysis on the second target type event sequence to determine the maximum mutual correlation coefficient corresponding to the second target type event sequence, the method further includes:
[0030] If the difference between the maximum mutual correlation coefficient corresponding to the second target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is not less than a preset difference, before each target type event in the second target type event sequence, respectively add the first type event under the corresponding preset interception time length to obtain a third target type event sequence;
[0031] If the difference between the maximum mutual correlation coefficient corresponding to the third target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is still not less than the preset difference, the third target type event sequence is used as the second target type event, and the step of adding the first type event under the corresponding preset interception time length before each target type event in the second target type event sequence to obtain the third target type event sequence is continued until the difference between the maximum mutual correlation coefficient corresponding to the third target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is less than the preset difference;
[0032] The third target type event sequence is used as the updated target type event sequence.
[0033] In an optional implementation manner, after determining the event type to which the precipitation data in the current time period belongs based on the precipitation data in the current time period and the water level data of the urban drainage system, the method further includes:
[0034] If the event type to which the precipitation data in the current time period belongs is a non-target type event, the precipitation data in the current time period is input into a flood prediction model corresponding to the non-target type event, and after being processed by the flood prediction model corresponding to the non-target type event, the water level data of the urban drainage system in the target area in the prediction time period is output;
[0035] Among them, the flood prediction model corresponding to the non-target type event is trained based on the precipitation data and water level data of the urban drainage system corresponding to multiple non-target type events in the target area within a preset time period, and the water level data of the urban drainage system in the target area during the prediction time period is used to characterize the risk of flooding in the target area during the prediction time period.
[0036] In a second aspect, the present disclosure provides a flood risk prediction device, the device comprising:
[0037] The first acquisition module is used to acquire the precipitation data of the target area in the current time period and the water level data of the urban drainage system;
[0038] A first determination module, configured to determine the event type to which the precipitation data in the current time period belongs based on the precipitation data in the current time period and the water level data of the urban drainage system;
[0039] A processing module, for inputting the precipitation data in the current time period into a flood prediction model corresponding to the target type event when the event type to which the precipitation data in the current time period belongs is a target type event, and outputting the water level data of the urban drainage system of the target area in the prediction time period after being processed by the flood prediction model corresponding to the target type event; wherein the flood prediction model corresponding to the target type event is trained based on the precipitation data and the water level data of the urban drainage system corresponding to a plurality of target type events in the target area within a preset time period, the target type event is used to characterize the event in which the water level in the urban drainage system rises due to precipitation, and the water level change in the urban drainage system exceeds a first preset threshold, and the water level data of the urban drainage system in the target area in the prediction time period is used to characterize the risk of flooding in the target area in the prediction time period.
[0040] In a third aspect, the present disclosure provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, and when the instructions are executed on a terminal device, the terminal device implements the above method.
[0041] In a fourth aspect, the present disclosure provides a flood risk prediction device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method when executing the computer program.
[0042] In a fifth aspect, the present disclosure provides a computer program product, wherein the computer program product comprises a computer program / instructions, and the computer program / instructions implement the above method when executed by a processor.
[0043] Compared with the prior art, the technical solution provided by the embodiments of the present disclosure has at least the following advantages:
[0044] The disclosed embodiments provide a flood risk prediction method, which improves the accuracy of flood risk prediction by performing flood prediction on the precipitation data of the target area in the current time period, based on the precipitation data corresponding to multiple target type events in the target area within a preset time period and the water level data of the urban drainage system to obtain a flood prediction model corresponding to the target type event. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.
[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.
[0047] Figure 1 A flow chart of a flood risk prediction method provided by an embodiment of the present disclosure;
[0048] Figure 2 A schematic diagram of precipitation event classification provided by an embodiment of the present disclosure;
[0049] Figure 3 A schematic diagram of an adjacent target type event provided by an embodiment of the present disclosure;
[0050] Figure 4 A schematic diagram of another adjacent target type event provided by an embodiment of the present disclosure;
[0051] Figure 5 A schematic diagram of another adjacent target type event provided by an embodiment of the present disclosure;
[0052] Figure 6 A flow chart of another flood risk prediction method provided by an embodiment of the present disclosure;
[0053] Figure 7 A schematic diagram of the structure of a flood risk prediction device provided by an embodiment of the present disclosure;
[0054] Figure 8 A schematic diagram of the structure of a flood risk prediction device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION
[0055] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.
[0056] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.
[0057] Urban flood disasters have an important impact on economic and social development. As the impact of climate change intensifies, extreme rainfall events in cities may further intensify, further affecting the sustainable development of urban economic and social development. Among them, urban floods can be described as the overflow of drainage systems in urban areas, which are generally caused by high-intensity rainfall, intensive surface runoff or insufficient drainage capacity.
[0058] Urban flood processes can be simulated and predicted using physical models or urban flood real-time prediction numerical models. Due to the limitations of physical models such as high data requirements and the need for consistency of physical structures, numerical models are more widely used in practice. Urban flood real-time prediction numerical models are usually based on time series artificial neural networks and machine learning for prediction. Compared with physical models, artificial neural networks and machine learning models have the advantages of relatively low model data parameter requirements, higher flexibility, and higher computational efficiency. These advantages enable the model to be quickly modeled after each flood occurs, providing more accurate and timely predictions in a short period of time, helping decision makers to formulate emergency plans. Urban real-time flood prediction models can generally provide early warnings for flood control and emergency response, reduce the impact of flood disasters, and play a key role in the emergency planning and real-time operation of urban flood early warning systems.
[0059] Existing models mainly focus on flood susceptibility, urban drainage system overflow prediction, flood inundation, flood risk analysis and flood warning. Some models also focus on water level prediction of combined sewer systems or rainwater collection systems. However, machine learning models are black box models and it is difficult to accurately describe the process. Therefore, the explanation of the process is insufficient, which limits the practical application of the model. The prediction accuracy of different forecast periods is still the main problem of existing models. The forecast period is the flood prediction period, which is defined as the delay between the real-time / current time step and the flood prediction time step. Specifically, the existing models have good prediction accuracy for short forecast periods of less than one hour, and limited prediction accuracy for long forecast periods. For example, the accurate prediction time of water level is generally within 2 hours.
[0060] On the one hand, monitoring data is another important factor affecting the accuracy of model simulation. Long-sequence monitoring data of rainfall and floods in urban drainage systems are also important for the development of simulation models, especially for the calibration and verification of individual flood events. Due to the lack of accurate definition and classification of urban flood events in the past, there are deviations in the calibration and verification of models that affect the accuracy. Distinguishing between flood events and non-flood events in urban drainage systems is conducive to improving model accuracy, but accurately defining flood events is the main problem at present.
[0061] On the other hand, traditional real-time flood prediction models rely on continuous time series data. Generally, real-time flood prediction models of urban drainage systems using nonlinear function regression methods are used to simulate and predict rainfall and flood events. They do not distinguish between different parts of the entire rainfall and flood event, and are interspersed with a lot of invalid data. The models are often more complex, require a lot of calculations, and are time-consuming, and usually have limited accuracy.
[0062] In summary, in order to improve the accuracy of flood risk prediction, the embodiments of the present disclosure provide a flood risk prediction method.
[0063] Specifically, precipitation data of the target area in the current time period and water level data of the urban drainage system are obtained, and based on the precipitation data in the current time period and the water level data of the urban drainage system, the event type to which the precipitation data in the current time period belongs is determined; if the event type to which the precipitation data in the current time period belongs is a target type event, the precipitation data in the current time period is input into a flood prediction model corresponding to the target type event; after being processed by the flood prediction model corresponding to the target type event, the water level data of the urban drainage system in the target area in the prediction time period is output; wherein the flood prediction model corresponding to the target type event is trained based on precipitation data and water level data of the urban drainage system corresponding to a plurality of target type events in the target area within a preset time period; the target type event is used to characterize an event in which the water level in the urban drainage system rises due to precipitation and the water level change in the urban drainage system exceeds a first preset threshold; and the water level data of the urban drainage system in the target area in the prediction time period is used to characterize the risk of flooding in the target area in the prediction time period. It can be seen that the embodiment of the present disclosure performs flood risk prediction on the precipitation data of the target area in the current time period by training a flood prediction model corresponding to the target type event based on the precipitation data corresponding to multiple target type events in the target area within a preset time period and the water level data of the urban drainage system, thereby improving the accuracy of flood risk prediction.
[0064] Based on this, the present disclosure provides a flood risk prediction method, referring to Figure 1 , is a flow chart of a flood risk prediction method provided by an embodiment of the present disclosure, the method comprising:
[0065] S101: Obtain precipitation data of a target area in a current time period and water level data of a city drainage system.
[0066] The flood risk prediction method provided by the embodiments of the present disclosure can be applied to the risk prediction of urban floods. Specifically, urban floods can be described as overflows of urban drainage systems due to precipitation.
[0067] In the disclosed embodiment, the target area may be any area where flood risk prediction is required.
[0068] In the disclosed embodiments, precipitation data refers to data recording the precipitation process. For example, the precipitation data may be rainfall data. Specifically, the rainfall data may represent the rainfall values corresponding to each time point within a certain time period. The water level data of the urban drainage system refers to data related to the water level height in facilities such as pipes, sewers, and drainage channels.
[0069] S102: Based on the precipitation data in the current time period and the water level data of the urban drainage system, determine the event type to which the precipitation data in the current time period belongs.
[0070] In the disclosed embodiment, the precipitation conditions in the current time period are first determined based on the precipitation data in the current time period, and the water level changes in the urban drainage system are determined based on the water level data of the urban drainage system. Then, based on the precipitation conditions and the water level changes in the urban drainage system, the event type to which the precipitation data in the current time period belongs is determined.
[0071] In the disclosed embodiment, the water level change in the urban drainage system refers to the difference between the current water level and the base flow water level. The base flow water level refers to the stable water level of the urban drainage system in the absence of any external water input or intervention in the urban drainage system. When precipitation or other factors cause the water level to rise, the water level change is the difference between the current water level and the base flow water level.
[0072] Specifically, the first type of event is used to characterize an event in which there is no precipitation and the water level change in the urban drainage system does not exceed a first preset threshold.
[0073] In the embodiment of the present disclosure, the water level change in the urban drainage system not exceeding the first preset threshold value means that the water level in the urban drainage system has no change or has a slight change.
[0074] The second type of event is used to characterize an event in which there is no precipitation and the water level in the urban drainage system drops from the flood level to the base flow level.
[0075] The third type of event is used to characterize an event in which there is no precipitation, the water level in the urban drainage system rises, and the change in the water level in the urban drainage system exceeds a first preset threshold.
[0076] In the disclosed embodiment, the third type of event is used to characterize an event in which there is no precipitation, but runoff is formed due to sewage treatment and other reasons, causing the water level in the urban drainage system to rise, and the water level change in the urban drainage system exceeds the first preset threshold.
[0077] The fourth type of event is used to characterize an event in which there is precipitation and the water level change in the urban drainage system does not exceed the first preset threshold.
[0078] In the disclosed embodiment, the fourth type of event is used to characterize an event in which there is precipitation, but due to evaporation, infiltration, etc., no runoff is formed, and the water level in the urban drainage system does not rise significantly.
[0079] The target type event is used to characterize the rising of water level in the urban drainage system caused by precipitation, and the event that the water level change in the urban drainage system exceeds a first preset threshold.
[0080] The first preset threshold may be set based on demand, and the embodiment of the present disclosure does not impose any limitation thereto.
[0081] S103: If the event type to which the precipitation data in the current time period belongs is a target type event, the precipitation data in the current time period is input into the flood prediction model corresponding to the target type event. After being processed by the flood prediction model corresponding to the target type event, the water level data of the urban drainage system in the target area in the prediction time period is output.
[0082] Among them, the flood prediction model corresponding to the target type event is trained based on the precipitation data and water level data of the urban drainage system corresponding to multiple target type events in the target area within a preset time period. The target type event is used to characterize the rise in water level in the urban drainage system caused by precipitation, and the water level change in the urban drainage system exceeds the first preset threshold. The water level data of the urban drainage system in the target area in the prediction time period is used to characterize the risk of flooding in the target area in the prediction time period.
[0083] In the disclosed embodiment, since precipitation needs time to gradually accumulate before entering the urban drainage system, that is, there is a lag time between precipitation and water level rise, that is to say, after the precipitation starts, the water level of the urban drainage system will begin to rise after a certain lag time. For this reason, by inputting the precipitation data in the current time period into the flood prediction model corresponding to the target type event, after being processed by the flood prediction model corresponding to the target type event, the water level data of the urban drainage system in the target area in the prediction time period can be output.
[0084] The predicted time period may be the lag time corresponding to the current time period.
[0085] In the disclosed embodiment, after obtaining the precipitation data of the target area in the current time period and the water level data of the urban drainage system, the precipitation data in the current time period is processed by the flood prediction model corresponding to the target type event to output the water level data of the urban drainage system in the target area in the predicted time period. Since the flood prediction model corresponding to the target type event is trained based on the precipitation data and the water level data of the urban drainage system corresponding to multiple target type events in the target area within the preset time period, the flood risk of the target area can be predicted more accurately.
[0086] In an optional embodiment, the time period corresponding to the target type event includes the time point when the water level rises to exceed the flood level. The time period corresponding to the target type event starts at the time point when the water level in the urban drainage system rises due to precipitation and the water level change in the urban drainage system exceeds the first preset threshold value for the first time, and ends at the time point when the water level in the urban drainage system recedes to the flood level after the precipitation ends.
[0087] In the disclosed embodiment, the flood level refers to the critical water level in the urban drainage system caused by precipitation, which rises to a level that exceeds the design capacity of the urban drainage system. When the water level exceeds the flood level, the drainage system will not be able to effectively discharge water, causing water accumulation and possibly causing floods.
[0088] For ease of understanding, illustratively, refer to Figure 2 , is a schematic diagram of precipitation event classification provided by an embodiment of the present disclosure, such as Figure 2 The time period corresponding to the target type event includes the time point when the water level rises above the flood level, such as point A1. Specifically, the time period corresponding to the target type event starts at the time point when the water level in the urban drainage system rises due to precipitation and the water level change in the urban drainage system exceeds the first preset threshold for the first time, i.e. point A0, and ends at the time point when the water level in the urban drainage system recedes to the flood level after the precipitation ends, i.e. point A2.
[0089] In addition, before the precipitation data in the current time period is input into the flood prediction model corresponding to the target type event and processed by the flood prediction model corresponding to the target type event, and before the water level data of the urban drainage system in the target area in the prediction time period is output, it is also necessary to train the flood prediction model corresponding to the target type event. In an optional implementation, the precipitation data of the target area in a preset time period and the water level data of the urban drainage system are obtained, the precipitation data of the target area in the preset time period and the water level data of the urban drainage system are classified, and the precipitation data and the water level data of the urban drainage system corresponding to multiple target type events in the target area in the preset time period are determined. Model training is performed based on the precipitation data and the water level data of the urban drainage system corresponding to multiple target type events in the target area in the preset time period to obtain a flood prediction model corresponding to the target type event.
[0090] In the embodiment of the present disclosure, the preset time period refers to a time period corresponding to the preset precipitation data of the target area and the water level data of the urban drainage system.
[0091] Specifically, the precipitation data of the target area within a preset time period and the water level data of the urban drainage system can be obtained from historical monitoring data and real-time monitoring data.
[0092] In the disclosed embodiment, after obtaining the precipitation data of the target area within a preset time period and the water level data of the urban drainage system, the precipitation data of the target area within the preset time period and the water level data of the urban drainage system are classified based on the precipitation situation and the water level change in the urban drainage system, and the precipitation data and the water level data of the urban drainage system corresponding to multiple types of events in the target area within the preset time period are obtained. Then, from the precipitation data and the water level data of the urban drainage system corresponding to the multiple types of events, the precipitation data and the water level data of the urban drainage system corresponding to multiple target type events in the target area within the preset time period are determined, so as to perform model training based on the precipitation data and the water level data of the urban drainage system corresponding to the multiple target type events in the target area within the preset time period, and obtain a flood prediction model corresponding to the target type events.
[0093] Among them, the multiple types of events at least include multiple first type events, multiple second type events and multiple target type events. On this basis, the multiple types of events can also include multiple third type events and / or multiple fourth type events.
[0094] Specifically, please refer to the above description for the definition of each type of event.
[0095] In addition, in order to improve the accuracy of flood risk prediction, before classifying the precipitation data of the target area within the preset time period and the water level data of the urban drainage system, and determining the precipitation data and the water level data of the urban drainage system corresponding to multiple target type events in the target area within the preset time period, data checking and completion are required. In an optional implementation, if there is missing data in the precipitation data of the target area within the preset time period and / or the water level data of the urban drainage system, a preset interpolation function is used to interpolate the missing data to obtain interpolated data. Based on the interpolated data and the precipitation data of the target area within the preset time period and the water level data of the urban drainage system, the corrected data is determined, and the corrected data is used as the precipitation data of the target area within the preset time period and the water level data of the urban drainage system.
[0096] In the embodiment of the present disclosure, the preset difference function may also be a continuous regression function, and the embodiment of the present disclosure does not impose any limitation on this.
[0097] In the disclosed embodiment, missing data can be filled by interpolating the missing data using a preset interpolation function, so that the precipitation data of the target area within the preset time period and the water level data of the urban drainage system are continuous data, thereby ensuring the continuity of the data, so that the flood prediction model corresponding to the target type event obtained by training based on the precipitation data of the target area within the preset time period and the water level data of the urban drainage system corresponding to multiple target type events in the target area within the preset time period is more accurate, thereby improving the accuracy of flood risk prediction.
[0098] In an optional implementation, after classifying the precipitation data of the target area within a preset time period and the water level data of the urban drainage system, and determining the precipitation data and the water level data of the urban drainage system corresponding to multiple target type events in the target area within the preset time period, the precipitation data and the water level data corresponding to the multiple target type events in the target area within the preset time period are sorted according to time information to obtain a target type event sequence, and interpolation and smoothing processing are performed on each target type event in the target type event sequence to obtain an updated target type event sequence. Accordingly, the updated target type event sequence is used for model training to obtain a flood prediction model corresponding to the target type event.
[0099] In order to more clearly understand the embodiment of the present disclosure, the embodiment of the present disclosure first explains the reason for performing interpolation and smoothing processing on each target type event in the target type event sequence:
[0100] For ease of understanding, any adjacent first target type event and second target type event in a target type event sequence are taken as an example for introduction.
[0101] refer to Figure 3 , is a schematic diagram of an adjacent target type event provided by an embodiment of the present disclosure, such as Figure 3 As shown, there are two water level values (i.e., L1 and L2) corresponding to the time point where the first target type event and the second target type event intersect (i.e., point D), resulting in data inconsistency. In order to eliminate the data inconsistency, data filling can be performed by interpolation smoothing after the first target type event and / or before the second target type event to eliminate the inconsistency of the water level value corresponding to point D.
[0102] To this end, in an embodiment of the present disclosure, after obtaining a target type event sequence, each target type event in the target type event sequence is interpolated and smoothed to solve the problem of inconsistency in water level values at the time point where any adjacent target type events in the target type event sequence intersect.
[0103] Furthermore, in order to solve the problem of inconsistency in water level values at the time point where any adjacent target type events in the target type event sequence intersect, the first type event and the second type event can be used to perform interpolation and smoothing processing on each target type event in the target type event sequence. Before interpolation and smoothing processing is performed on each target type event in the target type event sequence to obtain an updated target type event sequence, in an optional implementation, it is also necessary to perform a correlation analysis on the target type event sequence to determine the maximum correlation coefficient corresponding to the target type event sequence.
[0104] Among them, the maximum correlation coefficient is used to characterize the correlation between precipitation data and water level data under a preset lag time.
[0105] In the embodiment of the present disclosure, the preset lag time is determined based on the corresponding region, that is, different regions correspond to different lag times.
[0106] Specifically, the maximum cross-correlation coefficient
[0107] Among them, C o is the maximum correlation coefficient when the lag time is C, C is the lag time, x(i) is the precipitation data corresponding to the i-th time point in the target type event sequence, y(i) is the water level data of the urban drainage system corresponding to the i-th time point in the target type event sequence, is the average value of precipitation data in the target type event sequence, is the average value of the water level data of the urban drainage system in the target type event sequence, and m is the total number of time points in the target type event sequence, where m is a positive integer.
[0108] Accordingly, interpolation and smoothing processing is performed on each target type event in the target type event sequence to obtain an updated target type event sequence, including steps 1 to 4.
[0109] Step 1: before each target type event in the target type event sequence, add a first type event of a corresponding preset interception time length to obtain a first target type event sequence.
[0110] Among them, the first type of event is used to characterize an event in which there is no precipitation and the water level change in the urban drainage system does not exceed a first preset threshold.
[0111] In the disclosed embodiment, since the first type of event is an event that characterizes the absence of precipitation and the water level change in the urban drainage system does not exceed the first preset threshold, that is, an event that has little impact on the water level change in the urban drainage system caused by the first type of event, the first type of event with the corresponding preset cutoff time length is added before each target type event in the target type event sequence, which has little impact on the training of the flood prediction model corresponding to the subsequent target type event.
[0112] In the embodiment of the present disclosure, before each target type event in the target type event sequence, adding a first type event with a corresponding preset interception time length means that before each target type event in the target type event sequence, adding a corresponding first type event, and the time length of each added first type event is the corresponding preset interception time length.
[0113] In the embodiment of the present disclosure, the first type event corresponding to each target type event in the target type event sequence may be added before the first type event which is closest to the corresponding target type event.
[0114] The first type event that is closest to the corresponding target type event means that the time interval between the first type event and the target type event is the shortest.
[0115] Exemplarily, taking the first target type event as an example, before the first target type event, the first type event that is closest to the first target type event may be added.
[0116] In an optional embodiment, the preset interception time lengths corresponding to each added first type event may be the same. Specifically, the preset interception time length may be the average of the time lengths corresponding to each first type event in the first type event sequence, wherein the first type event sequence is obtained by sorting precipitation data and water level data corresponding to multiple first type events in the target area within a preset time period according to time information.
[0117] In another implementation manner, the preset interception time lengths corresponding to the added first type events may be different, that is, the added first type events correspond to different preset interception time lengths.
[0118] Take the second target type event as an example, Figure 4 As shown, the preset interception length D corresponding to the first type event closest to the second target type event A =C+P-Idx u
[0119] Among them, D A is the preset interception length corresponding to the first type event that is closest to the second target type event, C is the lag time mentioned above, P is the lead time of the time point corresponding to the time when the water level in the second target type event rises to the preset water level threshold, and Idx_u is the time difference between the start of the second target type event and the time when the water level rises to the preset water level threshold.
[0120] The lead time corresponding to when the water level in the second target type event rises to the preset water level threshold can be obtained based on the product of the time length of the second target type event and the preset percentage value.
[0121] Step 2: After each target type event in the first target type event sequence, add a corresponding second type event under a preset extension time length to obtain a second target type event sequence.
[0122] Among them, the second type of event is used to characterize the event when there is no precipitation and the water level in the urban drainage system drops from the flood level to the base flow level.
[0123] In the embodiment of the present disclosure, adding a second type event with a corresponding preset extended time length after each target type event in the first target type event sequence means that after each target type event in the target type event sequence, a corresponding second type event is added, and the time length of each added second type event is the corresponding preset extended length.
[0124] In the embodiment of the present disclosure, the starting point of the time length of each added second type event may be used as the extension starting point.
[0125] In the embodiment of the present disclosure, a corresponding second type event is added after each target type event in the target type event sequence, and the second type event may be the second type event that is closest to the corresponding target type event.
[0126] The second type event that is closest to the corresponding target type event means that the time interval between the second type event and the target type event is the shortest.
[0127] In an optional embodiment, the preset extension lengths corresponding to the added second type events may be the same. Specifically, the preset extension length may be the average of the time lengths corresponding to the respective second type events in the second type event sequence, wherein the second type event sequence is obtained by sorting the precipitation data and water level data corresponding to the plurality of second type events in the target area within a preset time period according to time information.
[0128] In another optional implementation, the preset extension lengths corresponding to the added second type events may be different.
[0129] Continuing with the second target type event as an example, the preset extension length corresponding to the second type event closest to the second target type event can be the lead time corresponding to the time point when the water level in the second target type event rises to the preset water level threshold, that is, the above-mentioned P.
[0130] Step 3: Perform a correlation analysis on the second target type event sequence to determine the maximum correlation coefficient corresponding to the second target type event sequence.
[0131] In the embodiment of the present disclosure, the process of determining the maximum mutual correlation coefficient corresponding to the second target type event sequence is similar to the above-mentioned process of determining the maximum mutual correlation coefficient corresponding to the target type event sequence, and please refer to the above description for details.
[0132] Step 4: If the difference between the maximum mutual correlation coefficient corresponding to the second target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is less than a preset difference, the second target type event sequence is used as the updated target type event sequence.
[0133] The preset difference value may be determined based on demand, and the embodiments of the present disclosure do not impose any limitation thereto.
[0134] In an embodiment of the present disclosure, if the difference between the maximum mutual correlation coefficient corresponding to the second target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is less than a preset difference, it indicates that the second target type event sequence and the target type event sequence have a high similarity. Therefore, the second target type event sequence can be used as an updated target type event sequence, and then the updated target type event sequence can be used for model training to obtain a flood prediction model corresponding to the target type event.
[0135] Further, it is assumed that the second target type event sequence includes the updated first target type event and the updated second target type event, such as Figure 5As shown, the updated first target type event refers to adding a first type event with a corresponding preset interception time length before the first target type event, and adding a second type event with a corresponding preset extension length after the first target type event, and the updated second target type event refers to adding a first type event with a corresponding preset interception time length before the second target type event, and adding a second type event with a corresponding preset extension length after the second target type event.
[0136] If there are still jump points in the second target type event sequence, that is, the end water level of the updated first target type event is inconsistent with the initial water level of the updated second target type event, new water level data can be formed by difference method to make the end water level of the updated first target type event consistent with the initial water level of the updated second target type event.
[0137] In an optional implementation, if the difference between the maximum mutual correlation coefficient corresponding to the second target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is not less than a preset difference, then before each target type event in the second target type event sequence, the first type event under the corresponding preset interception time length is added respectively to obtain a third target type event sequence; if the difference between the maximum mutual correlation coefficient corresponding to the third target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is still not less than the preset difference, then the third target type event sequence is used as the second target type event, and the step of adding the first type event under the corresponding preset interception time length before each target type event in the second target type event sequence to obtain the third target type event sequence is continued, until the difference between the maximum mutual correlation coefficient corresponding to the third target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is less than the preset difference, and the third target type event sequence is used as the updated target type event sequence.
[0138] In the embodiment of the present disclosure, if the difference between the maximum mutual correlation coefficient corresponding to the second target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is not less than the preset difference, it indicates that the similarity between the second target type event sequence and the target type event sequence is low, and it is also necessary to add the first type event under the corresponding preset interception time length before each target type event in the second target type event sequence to obtain a third target type event sequence, perform correlation analysis on the third target type event sequence, determine the maximum mutual correlation coefficient corresponding to the third target type event sequence, and continue to determine whether the difference between the maximum mutual correlation coefficient corresponding to the third target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is less than the preset difference. If it is still not less than the preset difference, the third target type event sequence is used as the second target type event, and the step of adding the first type event under the corresponding preset interception time length before each target type event in the second target type event sequence to obtain the third target type event sequence is continued until the difference between the maximum mutual correlation coefficient corresponding to the third target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is less than the preset difference, and the third target type event sequence is used as the updated target type event sequence.
[0139] After obtaining the updated target type event sequence, the updated target type event sequence is used for model training to obtain the flood prediction model.
[0140] In the disclosed embodiment, the updated target type event sequence may be used to train a nonlinear autoregressive network model, thereby obtaining a flood prediction model corresponding to the target type event.
[0141] After step S102, in an optional implementation, if the event type to which the precipitation data in the current time period belongs is a non-target type event, the precipitation data in the current time period is input into a flood prediction model corresponding to the non-target type event, and after being processed by the flood prediction model corresponding to the non-target type event, the water level data of the urban drainage system in the target area in the prediction time period is output.
[0142] Among them, the flood prediction model corresponding to non-target type events is trained based on the precipitation data and water level data of the urban drainage system corresponding to multiple non-target type events in the target area within a preset time period. The water level data of the urban drainage system in the target area during the prediction time period is used to characterize the risk of flooding in the target area during the prediction time period.
[0143] In the embodiment of the present disclosure, the non-target type event refers to any event of the first type event, the second type event, the third type event, and the fourth type event mentioned above.
[0144] Specifically, please refer to the above description for the definition of each type of event.
[0145] Taking the non-target type event as the first type event as an example, if the event type of the precipitation data in the current time period is the first type event, the precipitation data in the current time period is input into the flood prediction model corresponding to the first type event. After being processed by the flood prediction model corresponding to the first type event, the water level data of the urban drainage system in the target area in the prediction time period is output.
[0146] Exemplarily, the flood prediction model corresponding to the first type of event can be obtained by training a nonlinear regression model based on precipitation data corresponding to multiple first target type events in the target area within a preset time period and water level data of the urban drainage system.
[0147] The flood prediction model corresponding to the second type of event can be obtained by training a feedforward neural network model based on precipitation data corresponding to multiple second target type events in the target area within a preset time period and water level data of the urban drainage system.
[0148] The flood prediction model corresponding to the third type of event can be obtained by training a nonlinear regression model based on precipitation data corresponding to multiple third target type events in the target area within a preset time period and water level data of the urban drainage system.
[0149] The flood prediction model corresponding to the fourth type of event can be obtained by training a nonlinear regression model based on precipitation data corresponding to multiple fourth target type events in the target area within a preset time period and water level data of the urban drainage system.
[0150] In the flood risk prediction method provided by the embodiment of the present disclosure, precipitation data of the target area in the current time period and water level data of the urban drainage system are obtained. Based on the precipitation data in the current time period and the water level data of the urban drainage system, the event type to which the precipitation data in the current time period belongs is determined. If the event type to which the precipitation data in the current time period belongs is a target type event, the precipitation data in the current time period is input into a flood prediction model corresponding to the target type event. After being processed by the flood prediction model corresponding to the target type event, the water level data of the urban drainage system in the target area in the prediction time period is output, wherein the flood prediction model corresponding to the target type event is trained based on precipitation data and water level data of the urban drainage system corresponding to a plurality of target type events in the target area within a preset time period. The target type event is used to characterize an event in which the water level in the urban drainage system rises due to precipitation and the water level change in the urban drainage system exceeds a first preset threshold value. The water level data of the urban drainage system in the target area in the prediction time period is used to characterize the risk of flooding in the target area in the prediction time period. It can be seen that the embodiment of the present disclosure performs flood risk prediction on the precipitation data of the target area in the current time period by training a flood prediction model corresponding to the target type event based on the precipitation data corresponding to multiple target type events in the target area within a preset time period and the water level data of the urban drainage system, thereby improving the accuracy of flood risk prediction.
[0151] Based on the above method embodiment, the present disclosure also provides a flood risk prediction method, referring to Figure 6 , which is a flow chart of another flood risk prediction method provided in an embodiment of the present disclosure.
[0152] For ease of understanding, precipitation data is taken as an example, which is rainfall data caused by rainfall.
[0153] S601: Data collection.
[0154] In the disclosed embodiment, data collection refers to obtaining rainfall data of the target area within a preset time period and water level data of the urban drainage system. Specifically, the rainfall data of the target area within a preset time period and water level data of the urban drainage system can be obtained from historical monitoring data and real-time monitoring data.
[0155] S602: Event-based data identification.
[0156] The event-based data identification includes S602 - 1 event identification and S602 - 1 event-based data set generation.
[0157] Specifically, S602-1 event-based data identification includes data checking and completion, rainfall runoff event identification, and rainfall runoff event data storage.
[0158] Specifically, data checking and completion refers to analyzing the rainfall data of the target area within a preset time period and the water level data of the urban drainage system collected by S601, identifying the missing rainfall data and / or water level data, and then using a continuous regression function to interpolate the data, that is, to supplement the missing data in S601. If there is no missing data in the data collected by S601, no supplement is required.
[0159] Rainfall runoff event identification refers to classifying the rainfall data of the target area within a preset time period and the water level data of the urban drainage system, and determining the rainfall data and water level data of the urban drainage system corresponding to multiple types of events in the target area within the preset time period.
[0160] The multiple types of events include multiple first type events, multiple second type events, multiple third type events, multiple fourth type events and multiple target type events.
[0161] Specifically, for the definitions of the first type event, the second type event, the third type event, the fourth type event, and the target type event, please refer to the above description.
[0162] In the embodiment of the present disclosure, the risk of flooding in the urban drainage system is defined as the probability that the water level in the drainage system due to rainfall rises above the flood level. A flood event occurs when the water level in the urban drainage system exceeds the flood level. Therefore, the embodiment of the present disclosure divides rainfall runoff events into five categories (i.e., the above-mentioned first type of event, second type of event, third type of event, fourth type of event, and target type of event) based on rainfall data and water level data in the urban drainage system.
[0163] Among the above five types of events, target type events are more likely to cause floods. Therefore, the rainfall data corresponding to the target type event set in the target area within a preset time period and the water level data of the urban drainage system can be used as training data to train the flood prediction model corresponding to the target type events.
[0164] The storage of rainfall runoff event data means that after classifying the rainfall data of the target area within a preset time period and the water level data of the urban drainage system, the rainfall data corresponding to multiple types of events in the target area within the preset time period and the water level data of the urban drainage system are stored in independent databases respectively. The data can be stored in the form of a matrix data set, where the columns of the matrix can represent each time point within the preset time period, and the rows of the matrix can represent the rainfall data and the water level data.
[0165] S602-1 Generate event-based data set.
[0166] In the disclosed embodiment, event-based dataset generation refers to the generation of a continuous rainfall and flood dataset, namely, the updated target type event sequence mentioned above.
[0167] Since the machine learning model requires continuous time series data for model training and verification, but the event data identified by S602-1 includes data of different time lengths and different rainfall runoff types (i.e., the first type of events, the second type of events, the third type of events, the fourth type of events, and the target type of events mentioned above), which affects the simulation effect in the machine learning model, the embodiment of the present disclosure introduces two parameters, lead time P and lag time C, for constructing a rainfall flood time data set (i.e., the updated target type event sequence mentioned above).
[0168] Specifically, the process of constructing the updated target type event sequence is described in the above embodiment, and the embodiment of the present disclosure will not be described in detail here.
[0169] S603: Machine learning model construction.
[0170] In the disclosed embodiment, different types of events identified by S602, namely the first type of event, the second type of event, the third type of event, the fourth type of event, and the target type of event, are used to perform flood prediction for different types of rainfall events using different flood prediction models.
[0171] For first type events, flood prediction can be performed by training a nonlinear regression model using rainfall data corresponding to multiple first type events in the target area within a preset time period and water level data of the urban drainage system (i.e., a flood prediction model corresponding to the first type events).
[0172] For the third type of events, flood prediction can be performed by training a nonlinear regression model using rainfall data corresponding to multiple third type events in the target area within a preset time period and water level data of the urban drainage system (i.e., a flood prediction model corresponding to the third type of events).
[0173] For the fourth type of events, flood prediction can be performed by training a nonlinear regression model using rainfall data corresponding to multiple fourth type of events in the target area within a preset time period and water level data of the urban drainage system (i.e., a flood prediction model corresponding to the fourth type of events).
[0174] For target type events, flood prediction can be performed using a model obtained by training a nonlinear autoregressive network model using the above updated target type event sequence (ie, a flood prediction model corresponding to the target type event).
[0175] For the second type of events, flood prediction can be performed by training a feedforward neural network model using rainfall data corresponding to multiple second type events in the target area within a preset time period and water level data of the urban drainage system (i.e., a flood prediction model corresponding to the second type of events).
[0176] Specifically, different flood prediction models can be used to predict floods for different types of events.
[0177] First, obtain the real-time data corresponding to the target area, such as Dt (i.e., the rainfall data and urban system water level data corresponding to time t), and determine the event type corresponding to the real-time data Dt based on whether the water level rises and whether there is rainfall.
[0178] Exemplarily, if the Dt event is a target type event, water level prediction may be performed based on a flood prediction model corresponding to the target type event.
[0179] It should be noted that in step S602, the rainfall data corresponding to multiple types of events in the target area within a preset time period and the water level data of the urban drainage system can be divided into three parts, 70% for model training, 15% for verification, and 15% for model testing. The data set used for testing is also classified based on different rainfall event types (i.e., the above-mentioned first type of event, second type of event, third type of event, fourth type of event, and target type of event).
[0180] S604: Simulation effect evaluation.
[0181] Specifically, three indicators can be used for evaluation, namely, Nash efficiency coefficient, root mean square error and flood prediction accuracy.
[0182]
[0183] Where NSE is the Nash efficiency coefficient, n is the total number of simulation values; Y i is the i-th water level monitoring value; Y pi is the ith water level simulation value corresponding to the ith water level monitoring value.
[0184]
[0185] Where RMSE is the root mean square error, n is the total number of simulated values; Y i is the i-th water level monitoring value; Y pi is the ith water level simulation value corresponding to the ith water level monitoring value.
[0186]
[0187] Among them, FFA is the flood prediction accuracy, that is, the ratio of the number of accurately predicted floods to the total number of floods, TP is the number of accurately predicted floods, TN is the number of accurately predicted non-floods, FP is the number of floods that were not accurately predicted, and FN is the number of non-floods that were not accurately predicted.
[0188] At the same time, the embodiment of the present disclosure uses the relative increase of the root mean square error and the relative error to perform model sensitivity analysis, and the main analysis contents include:
[0189] (1) Sensitivity analysis of model parameters, including delay factor (the range of input rainfall data for each training iteration), feedback factor (the range of water level data for each training iteration), layers, nodes, and random training seeds.
[0190] (2) The accuracy of the model in predicting water levels for different rainfall event types (i.e., the first type of event, the second type of event, the third type of event, the fourth type of event, and the target type of event mentioned above).
[0191] (3) Uncertainty analysis of the model’s simulation effect on training data sets of different sizes.
[0192] (4) Model performance based on different rainfall characteristics, such as intensity and duration.
[0193]
[0194] Among them, ReRMSEI is the relative increase of root mean square error;
[0195] RMSE initial is the root mean square error of the model;
[0196] RMSE new is the new root mean square error of the model.
[0197]
[0198] Where Re is the relative error;
[0199] Y i is the i-th water level monitoring value;
[0200] Y pi is the ith water level simulation value corresponding to the ith water level monitoring value.
[0201] The test results show that compared with the traditional flood prediction model, the flood prediction model constructed by the embodiment of the present disclosure significantly improves the accuracy of flood prediction. The prediction accuracy for the 15-minute forecast period is 91%, the prediction accuracy for the 1-hour forecast period is 77%, and the prediction accuracy for the 3-hour forecast period is 53%.
[0202] In the disclosed embodiment, by classifying rainfall runoff events, different types of events are simulated and predicted using different methods, which can better identify the characteristics of different events and improve the accuracy and efficiency of simulation and prediction. Compared with traditional models, the disclosed embodiment has higher accuracy in predicting water level rise, especially in flood prediction with a longer forecast period of more than 2 hours. The present invention can reduce the root mean square error by 50%, improve flood prediction accuracy by 50%, and improve the Nash efficiency coefficient by 20%.
[0203] Based on the above method embodiment, the present disclosure also provides a flood risk prediction device, referring to Figure 7 , is a schematic diagram of a flood risk prediction device provided by an embodiment of the present disclosure, the device comprising:
[0204] The first acquisition module 701 is used to acquire precipitation data of the target area in the current time period and water level data of the urban drainage system;
[0205] A first determination module 702 is used to determine the event type to which the precipitation data in the current time period belongs based on the precipitation data in the current time period and the water level data of the urban drainage system;
[0206] Processing module 703 is used for inputting the precipitation data in the current time period into a flood prediction model corresponding to the target type event when the event type to which the precipitation data in the current time period belongs is a target type event, and outputting the water level data of the urban drainage system of the target area in the prediction time period after being processed by the flood prediction model corresponding to the target type event; wherein the flood prediction model corresponding to the target type event is trained based on the precipitation data and the water level data of the urban drainage system corresponding to a plurality of target type events in the target area within a preset time period, the target type event is used to characterize the event in which the water level rise in the urban drainage system caused by precipitation and the water level change in the urban drainage system exceeds a first preset threshold value, and the water level data of the urban drainage system of the target area in the prediction time period is used to characterize the risk of flooding in the target area in the prediction time period.
[0207] In an optional embodiment, the time period corresponding to the target type event includes the time point when the water level rises to exceed the flood level. The time period corresponding to the target type event starts at the time point when the water level rises in the urban drainage system caused by precipitation and the water level change in the urban drainage system first exceeds the first preset threshold, and ends at the time point when the water level in the urban drainage system recedes to the flood level after the precipitation ends.
[0208] In an optional implementation, the device further includes:
[0209] A second acquisition module is used to acquire precipitation data of the target area within a preset time period and water level data of the urban drainage system;
[0210] A classification module, used to classify the precipitation data of the target area within the preset time period and the water level data of the urban drainage system, and determine the precipitation data and the water level data of the urban drainage system corresponding to a plurality of target type events of the target area within the preset time period;
[0211] The first training module is used to perform model training based on precipitation data corresponding to multiple target type events in the target area within the preset time period and water level data of the urban drainage system, so as to obtain a flood prediction model corresponding to the target type event.
[0212] In an optional implementation, the device further includes:
[0213] An interpolation processing module, used for interpolating the missing data using a preset interpolation function to obtain interpolated data when there is missing data in the precipitation data of the target area within the preset time period and / or the water level data of the urban drainage system;
[0214] The second determination module is used to determine the corrected data based on the interpolation data and the precipitation data of the target area within the preset time period and the water level data of the urban drainage system, and use the corrected data as the precipitation data of the target area within the preset time period and the water level data of the urban drainage system.
[0215] In an optional implementation, the device further includes:
[0216] A sorting module, used to sort the precipitation data and water level data corresponding to the multiple target type events in the target area within the preset time period according to time information to obtain a target type event sequence;
[0217] An interpolation smoothing processing module, used for performing interpolation smoothing processing on each target type event in the target type event sequence to obtain an updated target type event sequence;
[0218] Accordingly, the first training module is specifically used for:
[0219] The updated target type event sequence is used to perform model training to obtain a flood prediction model corresponding to the target type event.
[0220] In an optional implementation, the device further includes:
[0221] A third determination module is used to perform a correlation analysis on the target type event sequence to determine a maximum mutual correlation coefficient corresponding to the target type event sequence; wherein the maximum mutual correlation coefficient is used to characterize the degree of correlation between the precipitation data and the water level data under a preset lag time;
[0222] Accordingly, the interpolation smoothing processing module includes:
[0223] A first adding submodule is used to add first type events of the corresponding preset interception time length before each target type event in the target type event sequence to obtain a first target type event sequence; wherein the first type event is used to represent an event in which there is no precipitation and the water level change in the urban drainage system does not exceed a first preset threshold value;
[0224] The second adding submodule is used to add the corresponding second type events with preset extension time length after each target type event in the first target type event sequence to obtain a second target type event sequence; wherein the second type event is used to characterize an event in which there is no precipitation and the water level in the urban drainage system drops from the flood level to the base flow water level.
[0225] A first determination submodule is used to perform a correlation analysis on the second target type event sequence to determine a maximum mutual correlation coefficient corresponding to the second target type event sequence;
[0226] The second determining submodule is used to use the second target type event sequence as an updated target type event sequence when the difference between the maximum mutual correlation coefficient corresponding to the second target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is less than a preset difference.
[0227] In an optional implementation, the device further includes:
[0228] A first adding module is used for adding the first type event with the corresponding preset interception time length before each target type event in the second target type event sequence when the difference between the maximum mutual correlation coefficient corresponding to the second target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is not less than a preset difference, to obtain a third target type event sequence;
[0229] a second adding module, for, when the difference between the maximum mutual correlation coefficient corresponding to the third target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is still not less than the preset difference, taking the third target type event sequence as the second target type event, and continuing to perform the step of adding the first type event under the corresponding preset interception time length before each target type event in the second target type event sequence to obtain a third target type event sequence, until the difference between the maximum mutual correlation coefficient corresponding to the third target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is less than the preset difference;
[0230] The fourth determining module is used to use the third target type event sequence as an updated target type event sequence.
[0231] In an optional implementation, the device further includes:
[0232] The second training module is used for inputting the precipitation data in the current time period into the flood prediction model corresponding to the non-target type event when the event type to which the precipitation data in the current time period belongs is a non-target type event, and outputting the water level data of the urban drainage system in the target area in the prediction time period after being processed by the flood prediction model corresponding to the non-target type event;
[0233] Among them, the flood prediction model corresponding to the non-target type event is trained based on the precipitation data and water level data of the urban drainage system corresponding to multiple non-target type events in the target area within a preset time period, and the water level data of the urban drainage system in the target area during the prediction time period is used to characterize the risk of flooding in the target area during the prediction time period.
[0234] In the flood risk prediction method provided by the embodiment of the present disclosure, precipitation data of the target area in the current time period and water level data of the urban drainage system are obtained. Based on the precipitation data in the current time period and the water level data of the urban drainage system, the event type to which the precipitation data in the current time period belongs is determined. If the event type to which the precipitation data in the current time period belongs is a target type event, the precipitation data in the current time period is input into a flood prediction model corresponding to the target type event. After being processed by the flood prediction model corresponding to the target type event, the water level data of the urban drainage system in the target area in the prediction time period is output, wherein the flood prediction model corresponding to the target type event is trained based on precipitation data and water level data of the urban drainage system corresponding to a plurality of target type events in the target area within a preset time period. The target type event is used to characterize an event in which the water level in the urban drainage system rises due to precipitation and the water level change in the urban drainage system exceeds a first preset threshold value. The water level data of the urban drainage system in the target area in the prediction time period is used to characterize the risk of flooding in the target area in the prediction time period. It can be seen that the embodiment of the present disclosure performs flood risk prediction on the precipitation data of the target area in the current time period by training a flood prediction model corresponding to the target type event based on the precipitation data corresponding to multiple target type events in the target area within a preset time period and the water level data of the urban drainage system, thereby improving the accuracy of flood risk prediction.
[0235] In addition to the above-mentioned method and apparatus, the embodiments of the present disclosure further provide a computer-readable storage medium, in which instructions are stored. When the instructions are executed on a terminal device, the terminal device implements the flood risk prediction method described in the embodiments of the present disclosure.
[0236] The embodiment of the present disclosure further provides a computer program product, which includes a computer program / instructions. When the computer program / instructions are executed by a processor, the flood risk prediction method described in the embodiment of the present disclosure is implemented.
[0237] In addition, the present disclosure also provides a flood risk prediction device, see Figure 8 As shown, this may include:
[0238] Processor 801, memory 802, input device 803 and output device 804. The number of processors 801 in the flood risk prediction device can be one or more. Figure 8 In some embodiments of the present disclosure, the processor 801, the memory 802, the input device 803 and the output device 804 may be connected via a bus or other means, wherein: Figure 8 The example of connecting through bus is taken in the following.
[0239] The memory 802 can be used to store software programs and modules. The processor 801 executes various functional applications and data processing of the flood risk prediction device by running the software programs and modules stored in the memory 802. The memory 802 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function, etc. In addition, the memory 802 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices. The input device 803 can be used to receive input digital or character information, and generate signal input related to user settings and function control of the flood risk prediction device.
[0240] Specifically in this embodiment, the processor 801 will load the executable files corresponding to the processes of one or more applications into the memory 802 according to the following instructions, and the processor 801 will run the applications stored in the memory 802, thereby realizing the various functions of the above-mentioned flood risk prediction device.
[0241] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.
[0242] The above description is only a specific embodiment of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A flood risk prediction method, characterized in that: The method comprises: Obtain precipitation data of the target area in the current time period and water level data of the urban drainage system; Determine the event type to which the precipitation data in the current time period belongs based on the precipitation data in the current time period and the water level data of the urban drainage system; If the event type to which the precipitation data in the current time period belongs is a target type event, the precipitation data in the current time period is input into a flood prediction model corresponding to the target type event, and after being processed by the flood prediction model corresponding to the target type event, the water level data of the urban drainage system of the target area in the prediction time period is output; wherein the flood prediction model corresponding to the target type event is obtained by training based on precipitation data and water level data of the urban drainage system corresponding to a plurality of target type events in the target area within a preset time period, the target type event is used to characterize an event in which the water level in the urban drainage system rises due to precipitation, and the water level change in the urban drainage system exceeds a first preset threshold, and the water level data of the urban drainage system of the target area in the prediction time period is used to characterize the risk of flooding in the target area in the prediction time period; If the event type to which the precipitation data in the current time period belongs is a non-target type event, the precipitation data in the current time period is input into a flood prediction model corresponding to the non-target type event, and after being processed by the flood prediction model corresponding to the non-target type event, the water level data of the urban drainage system in the target area in the prediction time period is output; Among them, the flood prediction model corresponding to the non-target type event is obtained by training based on the precipitation data and water level data of the urban drainage system corresponding to multiple non-target type events in the target area within a preset time period. The water level data of the urban drainage system in the target area in the prediction time period is used to characterize the risk of flooding in the target area in the prediction time period. The non-target type event includes any type of event among the first type event, the second type event, the third type event and the fourth type event. The first type event is used to characterize an event in which there is no precipitation and the water level change in the urban drainage system does not exceed a first preset threshold value. The second type event is used to characterize an event in which there is no precipitation and the water level in the urban drainage system drops from the flood level to the base flow level. The third type event is used to characterize an event in which there is no precipitation, the water level in the urban drainage system rises, and the water level change in the urban drainage system exceeds the first preset threshold value. The fourth type event is used to characterize an event in which there is precipitation and the water level change in the urban drainage system does not exceed the first preset threshold value.
2. The method according to claim 1, characterized in that The time period corresponding to the target type event includes the time point when the water level rises to exceed the flood level. The time period corresponding to the target type event starts at the time point when the water level in the urban drainage system rises due to precipitation and the water level change in the urban drainage system exceeds the first preset threshold for the first time, and ends at the time point when the water level in the urban drainage system recedes to the flood level after the precipitation ends.
3. The method according to claim 1 or 2, characterized in that: The method further comprises: inputting the precipitation data in the current time period into the flood prediction model corresponding to the target type event, and outputting the water level data of the urban drainage system in the target area in the prediction time period after the precipitation data in the current time period is processed by the flood prediction model corresponding to the target type event. Obtaining precipitation data of the target area within a preset time period and water level data of the urban drainage system; Classifying the precipitation data of the target area within the preset time period and the water level data of the urban drainage system, and determining the precipitation data and the water level data of the urban drainage system corresponding to a plurality of target type events of the target area within the preset time period; Model training is performed based on precipitation data corresponding to a plurality of target type events in the target area within the preset time period and water level data of the urban drainage system to obtain a flood prediction model corresponding to the target type event.
4. The method according to claim 3, characterized in that Before classifying the precipitation data of the target area within the preset time period and the water level data of the urban drainage system, and determining the precipitation data and the water level data of the urban drainage system corresponding to the multiple target type events of the target area within the preset time period, the method further includes: If there are missing data in the precipitation data of the target area within the preset time period and / or the water level data of the urban drainage system, interpolation processing is performed on the missing data using a preset interpolation function to obtain interpolated data; Based on the interpolation data and the precipitation data of the target area within the preset time period and the water level data of the urban drainage system, the corrected data is determined and used as the precipitation data of the target area within the preset time period and the water level data of the urban drainage system.
5. The method according to claim 3, characterized in that: After classifying the precipitation data of the target area within the preset time period and the water level data of the urban drainage system, and determining the precipitation data and the water level data of the urban drainage system corresponding to a plurality of target type events in the target area within the preset time period, the method further includes: Sorting the precipitation data and water level data corresponding to the multiple target type events in the target area within the preset time period according to time information to obtain a target type event sequence; Performing interpolation and smoothing processing on each target type event in the target type event sequence to obtain an updated target type event sequence; Accordingly, the model training is performed based on the precipitation data and the water level data of the urban drainage system corresponding to the multiple target type events in the target area within the preset time period to obtain the flood prediction model corresponding to the target type event, including: The updated target type event sequence is used to perform model training to obtain a flood prediction model corresponding to the target type event.
6. The method according to claim 5, characterized in that Before performing interpolation and smoothing processing on each target type event in the target type event sequence to obtain an updated target type event sequence, the method further includes: Performing a correlation analysis on the target type event sequence to determine a maximum mutual correlation coefficient corresponding to the target type event sequence; wherein the maximum mutual correlation coefficient is used to characterize the degree of correlation between the precipitation data and the water level data under a preset lag time; Accordingly, performing interpolation and smoothing processing on each target type event in the target type event sequence to obtain an updated target type event sequence includes: Before each target type event in the target type event sequence, a first type event under a corresponding preset interception time length is added to obtain a first target type event sequence; wherein the first type event is used to characterize an event in which there is no precipitation and the water level change in the urban drainage system does not exceed a first preset threshold value; After each target type event in the first target type event sequence, a second type event with a corresponding preset extension time length is added to obtain a second target type event sequence; wherein the second type event is used to represent an event in which there is no precipitation and the water level in the urban drainage system drops from the flood level to the base flow level; performing a correlation analysis on the second target type event sequence to determine a maximum mutual correlation coefficient corresponding to the second target type event sequence; If the difference between the maximum mutual correlation coefficient corresponding to the second target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is less than a preset difference, the second target type event sequence is used as an updated target type event sequence.
7. The method according to claim 6, characterized in that After performing correlation analysis on the second target type event sequence to determine the maximum mutual correlation coefficient corresponding to the second target type event sequence, the method further includes: If the difference between the maximum mutual correlation coefficient corresponding to the second target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is not less than a preset difference, before each target type event in the second target type event sequence, respectively add the first type event under the corresponding preset interception time length to obtain a third target type event sequence; If the difference between the maximum mutual correlation coefficient corresponding to the third target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is still not less than the preset difference, the third target type event sequence is used as the second target type event, and the step of adding the first type event under the corresponding preset interception time length before each target type event in the second target type event sequence to obtain the third target type event sequence is continued until the difference between the maximum mutual correlation coefficient corresponding to the third target type event sequence and the maximum mutual correlation coefficient corresponding to the target type event sequence is less than the preset difference; The third target type event sequence is used as the updated target type event sequence.
8. A flood risk prediction device, characterized in that: The device comprises: The first acquisition module is used to acquire the precipitation data of the target area in the current time period and the water level data of the urban drainage system; A first determination module, configured to determine the event type to which the precipitation data in the current time period belongs based on the precipitation data in the current time period and the water level data of the urban drainage system; A processing module, for inputting the precipitation data in the current time period into a flood prediction model corresponding to the target type event when the event type to which the precipitation data in the current time period belongs is a target type event, and outputting the water level data of the urban drainage system in the target area in the prediction time period after being processed by the flood prediction model corresponding to the target type event; wherein the flood prediction model corresponding to the target type event is obtained by training based on precipitation data and water level data of the urban drainage system corresponding to a plurality of target type events in the target area within a preset time period, the target type event is used to characterize an event in which the water level in the urban drainage system rises due to precipitation, and the water level change in the urban drainage system exceeds a first preset threshold, and the water level data of the urban drainage system in the target area in the prediction time period is used to characterize the risk of flooding in the target area in the prediction time period; The second training module is used for inputting the precipitation data in the current time period into the flood prediction model corresponding to the non-target type event when the event type to which the precipitation data in the current time period belongs is a non-target type event, and outputting the water level data of the urban drainage system in the target area in the prediction time period after being processed by the flood prediction model corresponding to the non-target type event; Among them, the flood prediction model corresponding to the non-target type event is obtained by training based on the precipitation data and water level data of the urban drainage system corresponding to multiple non-target type events in the target area within a preset time period. The water level data of the urban drainage system in the target area in the prediction time period is used to characterize the risk of flooding in the target area in the prediction time period. The non-target type event includes any type of event among the first type event, the second type event, the third type event and the fourth type event. The first type event is used to characterize an event in which there is no precipitation and the water level change in the urban drainage system does not exceed a first preset threshold value. The second type event is used to characterize an event in which there is no precipitation and the water level in the urban drainage system drops from the flood level to the base flow level. The third type event is used to characterize an event in which there is no precipitation, the water level in the urban drainage system rises, and the water level change in the urban drainage system exceeds the first preset threshold value. The fourth type event is used to characterize an event in which there is precipitation and the water level change in the urban drainage system does not exceed the first preset threshold value.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed on a terminal device, the terminal device implements the method according to any one of claims 1 to 7.
10. A flood risk prediction device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.
Citation Information
Patent Citations
River interval dynamic flood peak analysis early warning method, system, equipment and medium
CN113379098A
Waterlogging prediction method and system and storage medium
CN114254561A
Urban precipitation inversion method and device, computer equipment and storage medium
CN116311039A
Intelligent early warning method and system for urban flood control
CN117010726A
KR20200127091A
