An urban flood disaster monitoring and early warning system and method

Abstract

The present invention relates to the technical field of flood disaster monitoring, and specifically relates to an urban flood disaster monitoring and early warning system and method. By using rainfall, soil humidity, temperature, historical water level and wind speed as input data of a linear regression model, and obtaining prediction data therefrom, and by constructing a time series data set from the prediction data and performing differencing processing, based on the constructed time series data set, a flood hybrid prediction model jointly established by a SARIMA model and an ARIMAX model is introduced to predict the water level value in the city, thereby improving the prediction accuracy and stability of the prediction model for the water level value; and based on the predicted water level value, a risk assessment model is combined to comprehensively judge whether there is a flood disaster risk in the city, and when there is, the emergency plan of the city is clarified, and a polynomial regression model is constructed therefrom to further predict the damage degree of the city at a future moment, so as to provide auxiliary decision-making support for the monitoring and disposal of urban flood disasters.

Description

Technical Field

[0001] The present invention relates to the technical field of flood disaster monitoring, and particularly relates to an urban flood disaster monitoring and early warning system and method. Background Art

[0002] Urban flood disasters refer to the serious losses caused by the rise of water levels in urban areas due to rainfall, poor drainage or other factors. Such disasters not only pose threats to people's lives and property safety, but also have a huge impact on urban infrastructure and the ecological environment. With the intensification of global climate change and the acceleration of urbanization, the frequency and intensity of flood disasters have gradually increased.

[0003] Currently, parameters such as rainfall, wind speed, and humidity are usually combined, and a linear regression model is used to predict the risk level of urban floods. However, when analyzing the risk level of urban floods based on these parameters, the temporal characteristics of the parameter data, such as seasonal changes, are not considered, resulting in poor performance of the model in processing temporal data, thereby affecting the accuracy of judgment.

[0004] Moreover, the linear regression model only incorporates parameters such as rainfall, wind speed, and humidity into the model, and does not combine the inherent characteristics and external characteristics of temporal data to predict flood risks, further reducing the accuracy of predicting the flood risk level. Summary of the Invention

[0005] In view of the above-mentioned drawbacks of the prior art, the present invention provides an urban flood disaster monitoring and early warning system and method, which can effectively solve the problem of inaccurate prediction of urban flood disaster situations in the prior art.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions:

[0007] The present invention provides an urban flood disaster monitoring and early warning system and method, including:

[0008] A data collection and construction module, which is used to obtain the rainfall, soil humidity, temperature, historical water level and wind speed of the city, establish a linear regression model to predict water level data, and construct a time series data set. It also includes:

[0009] A flood risk prediction and assessment module, based on the time series data set, introducing the influence of seasonal factors and rainfall, soil humidity, and wind speed, and establishing a flood hybrid prediction model by combining the SARIMA model and the ARIMAX model to predict the water level value of the city. Among them:

[0010] Based on the predicted water level value, clarify the rainfall, soil humidity and siltation conditions of the city at the predicted moment to construct a risk assessment model, and judge whether the city is a flood disaster city through the risk assessment model and execute early warning. Among them:

[0011] If a warning is issued, obtain the generated emergency plan to clarify the completion degree of the emergency plan at a future moment, and combine the completion degree with rainfall, soil humidity, and siltation conditions to establish a polynomial regression model to predict the damage degree of flood-stricken cities.

[0012] Further, after constructing the time series dataset, perform a stationarity test on the time series dataset to determine whether to perform differencing operations, where the stationarity test is performed through the ADF test or the KPSS test.

[0013] Further, the algorithm expression of the SARIMA model is:

[0014]

[0015] In the formula, is the differenced water level observation value at time , that is, the change amplitude of the water level value, is the seasonal autoregressive polynomial, is the autoregressive polynomial, is the autoregressive order, is the seasonal differencing operation, is the number of seasonal differencing times, is the non-seasonal differencing operation, is the number of non-seasonal differencing times, is the seasonal moving average polynomial, is the non-seasonal moving average part, is the order of the seasonal moving average term, is the error term.

[0016] Further, the algorithm model expression of the ARIMAX model after differencing is:

[0017]

[0018] In the formula, is the exogenous variable, including rainfall, soil humidity, wind speed, is the coefficient of the exogenous variable, is the number of exogenous variables.

[0019] Further, the algorithm expression of the flood hybrid prediction model is:

[0020]

[0021] In the formula, is the differencing operation, is the number of seasonal differencing times, is the number of non-seasonal differencing times.

[0022] Further, after obtaining the observed values after differential processing , perform the restoration of the predicted value, and the algorithm expression is as follows:

[0023]

[0024] In the formula: is the water level value at the current moment, is the water level value at the

[0025] Further, the algorithm expression of the risk assessment model is:

[0026]

[0027] In the formula, is the flood risk value, is the rainfall at the prediction moment, is the soil moisture at the prediction moment, is the siltation condition at the prediction moment, , , and are the corresponding weight coefficients respectively.

[0028] Further, the method for determining the completion degree of the emergency plan is:

[0029] Obtain the number of emergency response personnel in the current emergency plan ;

[0030] Obtain the number of emergency equipment used in the emergency plan, and determine the proportion of emergency equipment that can continue to work normally ;

[0031] Obtain the number of emergency paths planned in the emergency plan , determine the distances between the emergency paths and the areas that need to be dealt with in the emergency plan respectively, and obtain the total distance of the sum of multiple distances and record it as the emergency distance ,

[0032] Clarify the characteristics of the fork roads in the emergency paths, and obtain the number of the characteristics of the fork roads in all emergency paths and record it as the total number of fork roads and record it as ;

[0033] Based on:

[0034]

[0035] In the formula, , , , and 5 are the corresponding weight coefficients respectively, is the number of calculation items, preset to 5, is the constant correction coefficient;

[0036] The completion coefficient is obtained , and by clarifying the preset completion interval where the completion coefficient is located, the corresponding completion is obtained .

[0037] Furthermore, the algorithm expression of the polynomial regression model is as follows:

[0038]

[0039] In the formula, is the predicted damage degree, ,... are all model parameters, is the completion;

[0040] Among them, by introducing ElasticNet to combine L1 regularization and L2 regularization to optimize the polynomial regression model, its loss function is:

[0041]

[0042] In the formula, is the true value, is the predicted value, is the regression coefficient of the jth feature, is the number of samples, is the feature vector, is the regularization parameter, used to control the strength of regularization, and are the strengths of L1 regularization and L2 regularization respectively.

[0043] Different from the above, the present invention also provides a flood disaster monitoring and early warning method, which is implemented according to the described urban flood disaster monitoring and early warning system.

[0044] The technical solution provided by the present invention has the following beneficial effects compared with the known prior art:

[0045] By using rainfall, soil humidity, temperature, historical water level and wind speed as the input data of the linear regression model, and obtaining the prediction data therefrom, and by constructing a time series data set with the prediction data and performing differential processing, based on the constructed time series data set, a flood hybrid prediction model established by introducing the SARIMA model and the ARIMAX model is used to predict the water level value in the city, thereby improving the prediction accuracy and stability of the prediction model for the water level value;

[0046] And based on the predicted water level value, combined with the risk assessment model, comprehensively judge whether there is a risk of urban flood disaster. When there is a risk, clarify the urban emergency plan, and thus construct a polynomial regression model to further predict the damage degree of the city at future moments, so as to provide auxiliary decision-making support for the monitoring and disposal of urban flood disasters. Description of the Drawings

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0048] Figure 1 It is a schematic diagram of the overall process of the present invention. Detailed Embodiments

[0049] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0050] The following further describes the present invention with reference to the embodiments.

[0051] Embodiment 1 (refer to Figure 1 ): A monitoring and early warning system for urban flood disasters includes at least:

[0052] A data acquisition and construction module for obtaining the impact data related to urban and flood risks in different urban areas over a period of time. The impact data includes: rainfall, soil humidity, temperature, historical water level, and wind speed of the city. Thus, a linear regression model is established by introducing the relationship between the impact data and the water level. The algorithm expression of the linear regression model is as follows:

[0053]

[0054] In the formula, is the predicted water level data, is the intercept term, is the rainfall, is the soil humidity, is the temperature, is the wind speed, is the historical water level, , , , and are the corresponding influence weights respectively, is the error term;

[0055] A period of time in the above can be monthly time, etc. (such as calculating backward 30 days from the current time as the time node to obtain the influence data under historical monthly time), and the average value of the influence data under monthly time is clearly obtained for establishing a linear regression model and predicting the current water level data.

[0056] Thus, the predicted water level data is combined with the historical water level data to form a time series data set. Among them, a stationarity test is performed on the time series data set to judge the differencing operation to ensure the stationarity of the data. The stationarity test can be passed by ADF test or KPSS test. Based on this, it is judged whether the differencing operation needs to be performed. If it is not stationary, the differencing operation needs to be performed. The time series data set can be pre-differenced by the first order to judge whether it is stationary. If it is stationary, stop; otherwise, continue. Thus, the number of differencing times is determined to obtain a stationary time series data set. The differencing operation is a well-known prior art, so this case will not elaborate on it.

[0057] The flood risk prediction and assessment module is used to construct a time series model through the time series data set after differencing. It should be noted that in this embodiment, considering the influence of seasonal factors and exogenous variables (rainfall, soil humidity, wind speed) on the time series model, the seasonal factors, rainfall, soil humidity, and wind speed (exogenous variables) are introduced into the time series model. The time series model includes SARIMA model and ARIMAX model. The SARIMA model can process seasonal time series data and can effectively identify and model seasonal patterns. The ARIMAX model can make the prediction more comprehensive by introducing exogenous variables such as rainfall, soil humidity, and wind speed. Through exogenous variables, ARIMAX can reveal the influence of different factors on the target variable (water level value) and enhance the interpretability of the model.

[0058] Furthermore, the algorithm model expression of the SARIMA model after differencing is:

[0059]

[0060] The algorithm model expression of the ARIMAX model after differencing is:

[0061]

[0062] Among them, is time The differential water level observation value, i.e., the change range of the water level value, is the seasonal autoregressive polynomial, representing the seasonal autoregressive part, is the autoregressive polynomial, is the autoregressive order, is the seasonal differencing operation, is the number of seasonal differencing times, is the non-seasonal differencing operation, is the number of non-seasonal differencing times, is the seasonal moving average polynomial, The non-seasonal moving average part, is the order of the seasonal moving average term, is the error term, is the exogenous variable, including rainfall, soil moisture, wind speed, is the coefficient of the exogenous variable, is the number of exogenous variables.

[0063] Based on the above time series model, in order to enable the time series model to provide more accurate prediction results for urban flood risk prediction in a complex environment by simultaneously considering seasonal changes and exogenous variables, that is, relatively improving the model's understanding and prediction ability of complex dynamic systems. Therefore, by fusing the SARIMA model and the ARIMAX model, a flood hybrid prediction model for urban flood disasters is obtained to achieve accurate prediction of urban flood disasters. Among them, the expression of the flood hybrid prediction model is specifically:

[0064]

[0065] In the formula, is the differencing operation, is the number of seasonal differencing times, is the number of non-seasonal differencing times, and the observed value after differencing processing is obtained After that, in order to obtain the actual water level value, the predicted value needs to be restored. The algorithm expression involved is as follows:

[0066]

[0067] In the formula: is the observed value (water level value) at the current moment, is the observed value at the

[0068] It should be noted that when predicting the water level value in the city at subsequent moments, the moment to be predicted can be substituted to predict the water level value in the city at subsequent times (the predicted water level value can be expressed as , according to the above integrated flood prediction model, when predicting the predicted water level value at a future time, the predicted rainfall at the future time needs to be substituted. , soil moisture (refer to the following text) and wind speed. Specifically, the predicted wind speed can be obtained by jointly establishing a linear regression model through temperature, air pressure, humidity, rainfall, terrain features, and the wind speed at the previous time, which will not be elaborated here).

[0069] Further, the predicted water level value is obtained by predicting the water level value at subsequent times , and a line risk assessment model is created therefrom to evaluate the flood risk in the city. The specific steps are as follows:

[0070] Based on the predicted water level value , the rainfall, soil moisture, and siltation condition of the drainage pipeline at the prediction time are synchronously obtained;

[0071] And a risk assessment model is constructed therefrom:

[0072]

[0073] In the formula, is the flood risk value, is the rainfall at the prediction time, is the soil moisture at the prediction time, is the siltation condition at the prediction time, , , and are the corresponding weight coefficients respectively. Among them,

[0074] For the rainfall at the prediction time, it can be predicted by constructing a linear regression model. Specifically, weather parameters such as temperature, humidity, wind speed, and air pressure within a certain period are obtained. Through Pearson correlation coefficient or heat map analysis, the correlation between rainfall and weather conditions is determined to identify the key features affecting rainfall, usually including temperature, humidity, and wind speed. Thus, a linear regression model can be constructed to predict the rainfall at a future time;

[0075] Similarly for soil moisture, by obtaining temperature, wind speed, soil type, volumetric water content, saturated water content, and rainfall within a certain period, the soil moisture at a future time can be predicted based on the linear regression model;

[0076] For the sedimentation situation at the prediction time, sedimentation data over a certain period in the past are obtained in advance, including rainfall, flow rate, velocity, soil humidity, diameter of the drainage pipe, and height of the sediment (the sedimentation data are all average values), and corresponding weight coefficients are assigned to construct a linear regression model for prediction, and then the corresponding rainfall, soil humidity, and sedimentation situation at the prediction time are obtained. Thus, the flood risk value at the prediction time can be calculated according to the weighted method. , therefore, according to the flood risk value judge the risk interval at the prediction time, and then, the flood disaster risk degree of different urban areas at the prediction time can be clarified. Generally speaking, the larger the flood risk value , the higher the risk degree. When the flood risk value exceeds the preset safety interval, it is determined that there will be a flood disaster risk at the prediction time. Thus, a warning signal is formed and output to the terminal device to execute the warning.

[0077] Furthermore, in this solution, by clarifying whether there is a flood disaster risk at the prediction time in different urban areas, the urban areas with flood disaster risks are divided and recorded as flood disaster cities. At the same time, the prediction time when the flood disaster cities trigger the flood disaster is obtained and recorded as the flood prediction time. Then, multiple different flood prediction times are obtained. A disaster city dataset is constructed through the flood disaster cities, and the flood disaster cities in the disaster city dataset are sorted according to the flood prediction time from early to late. The time values between the corresponding flood prediction time and the current time are determined and recorded as the response time (multiple response times are obtained). The corresponding emergency response departments for the flood disaster cities are determined, and the emergency response plans generated by each emergency response department for the flood disaster cities are clarified (generally speaking, the emergency response plans are preset in the database based on the response time and the predicted water level value for activation in case of flood risk). The completion coefficient of the emergency response plan at the response time is determined. , among which, for the completion coefficient , the judgment basis is as follows:

[0078] Obtain the number of emergency response personnel in the current emergency response plan. ;

[0079] Obtain the number of emergency equipment used in the emergency response plan, and at the same time determine the proportion of the emergency equipment that can continue to work normally. , obtained according to the ratio between the emergency equipment without failure and the total emergency equipment.

[0080] Obtain the number of emergency paths planned in the emergency response plan. , and then determine the distances between the emergency paths and the areas that need to be disposed of in the emergency response plan respectively. The sum of the multiple distances, the total distance, is recorded as the emergency distance. ,

[0081] Define the characteristics of the fork in the emergency path. For the defined fork characteristics, when the application of the emergency path is affected, it is beneficial for the emergency plan executors to change the original emergency path based on the fork, so as to facilitate the execution of the emergency plan. Thus, obtain the number of fork characteristics of all emergency paths and record it as the total number of forks (determine the total number) and record it as ;

[0082] Thus, according to:

[0083]

[0084] In the formula, , , , and 5 are the corresponding weight coefficients respectively, is the number of calculation items, preset to 5, is the constant correction coefficient, and thus obtain the completion coefficient , clarify the preset completion interval where the completion coefficient is located, that is, obtain the corresponding completion , and thus obtain the execution completion degree of the emergency plan under the corresponding response time , furthermore, combine the completion with the predicted rainfall , soil humidity , siltation situation Introduce a polynomial regression model to predict the damage degree of flood-stricken cities at future times. The specific polynomial regression model is as follows:

[0085]

[0086] In the formula, is the predicted damage degree, ,... are all model parameters.

[0087] The data analysis and optimization module is used to optimize the polynomial regression model using ElasticNet and combine L1 and L2 regularization. Among them, the polynomial regression model is optimized through L1 regularization and L2 regularization to effectively reduce the influence of unimportant features. Among them, the loss function of L1 regularization is:

[0088]

[0089] Limit the complexity of the model by introducing the L2 regularization term, and its loss function is:

[0090]

[0091] In the formula, is the true value (usually historical data, such as the actual damage degree of the city at a certain historical time point for actual observations), is the predicted value, is the regression coefficient of the j-th feature, is the number of samples, is the feature vector (4), is the regularization parameter, used to control the strength of regularization;

[0092] Therefore, ElasticNet is introduced to optimize the polynomial regression model by combining the advantages of L1 regularization and L2 regularization to maintain the stability of the model, and its loss function is:

[0093]

[0094] In the formula, and are the strengths of L1 regularization and L2 regularization respectively. Thus, the risk of overfitting is reduced while the prediction stability of the above polynomial regression model is improved.

[0095] Furthermore, based on the predicted damage degree of the flood disaster city obtained by the above polynomial regression model, the damage degree of different flood disaster cities at future times and under the implementation of the emergency plan is clarified, so that the control center can pre-plan the safety of the city according to the damage degree.

[0096] Different from the above, the present invention also provides a flood disaster monitoring and early warning method, which is implemented according to the described urban flood disaster monitoring and early warning system, that is, referring to the above urban flood disaster monitoring and early warning system, which will not be elaborated here.

[0097] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements will not make the essence of the corresponding technical solutions deviate from the protection scope of the technical solutions of the embodiments of the present invention.

Claims

1. An urban flood disaster monitoring and early warning system, comprising: The data collection and construction module is used to obtain the city's rainfall, soil moisture, temperature, historical water level and wind speed, establish a linear regression model to predict water level data, and construct a time series data set, which is characterized by also including: The flood risk prediction and assessment module, based on the time series data set, introduces seasonal factors and the influence of rainfall, soil moisture, and wind speed, and combines the SARIMA model and the ARIMAX model to establish a hybrid flood prediction model to predict the water level value of the city, including: Based on the predicted water level value, the rainfall, soil moisture and sedimentation conditions of the city at the time of prediction are clearly defined to build a risk assessment model. Through the risk assessment model, it is determined whether the city is a flood disaster city and an early warning is implemented, including: If an early warning is issued, the generated emergency plan is obtained to clarify the completion degree of the emergency plan at the future moment, and the completion degree is combined with rainfall, soil moisture and siltation to establish a polynomial regression model to predict the degree of damage to the city caused by flood disasters; The algorithm expression of the flood hybrid prediction model is: In the formula, For differential operation, is the number of seasonal differences, is the number of non-seasonal differences, is an autoregressive polynomial, is the autoregressive order, For time The differential water level observations, is the seasonal autoregressive polynomial, is the seasonal moving average polynomial, is the error term, Non-seasonal moving average component.

2. The urban flood disaster monitoring and early warning system according to claim 1 is characterized in that: After the time series data set is constructed, a stationarity test is performed on the time series data set to determine whether to perform a difference operation, wherein the stationarity test is performed through an ADF test or a KPSS test.

3. The urban flood disaster monitoring and early warning system according to claim 1 is characterized in that: The algorithm expression of the SARIMA model is: In the formula, For time The differential water level observation value, that is, the change range of the water level value, is the seasonal autoregressive polynomial, is an autoregressive polynomial, is the autoregressive order, is the seasonal difference operation, is the number of seasonal differences, is a non-seasonal differencing operation, is the number of non-seasonal differences, is the seasonal moving average polynomial, The non-seasonal moving average part, is the order of the seasonal moving average term, is the error term.

4. The urban flood disaster monitoring and early warning system according to claim 3 is characterized in that: The algorithm model expression of the ARIMAX model after differentiation is: In the formula, are exogenous variables, including rainfall, soil moisture, and wind speed. is the coefficient of the exogenous variable, is the number of exogenous variables.

5. The urban flood disaster monitoring and early warning system according to claim 1 is characterized in that: Get the observed value after difference processing After that, the predicted value is restored. The algorithm expression is as follows: Where: is the water level value at the current moment, for The water level value at a certain moment.

6. The urban flood disaster monitoring and early warning system according to claim 1 is characterized in that: The algorithm expression of the risk assessment model is: In the formula, is the flood risk value, To predict the amount of rainfall at a given time, To predict the soil moisture at the time, To predict the sedimentation situation at the time, , , as well as are the corresponding weight coefficients respectively.

7. The urban flood disaster monitoring and early warning system according to claim 6 is characterized in that: The method for determining the degree of completion of the emergency plan is as follows: Get the number of emergency response personnel in the current emergency plan ; Obtain the number of emergency equipment used in the emergency plan and determine the percentage of emergency equipment that can continue to perform normal work ; Get the number of emergency routes planned in the emergency plan , determine the distance between the emergency path and the area that needs to be dealt with in the emergency plan, obtain the total distance of multiple distances and record it as the emergency distance , Identify the fork features in the emergency path, obtain the number of fork features of all emergency paths and record it as the total fork and record it as ; in accordance with: In the formula, , , , as well as 5 are the corresponding weight coefficients, To calculate the number of items, is the constant correction factor; Get the completion coefficient , clarify the preset completion interval of the completion coefficient, and get the corresponding completion coefficient. .

8. The urban flood disaster monitoring and early warning system according to claim 6 is characterized in that: The algorithm expression of the polynomial regression model is as follows: In the formula, For the predicted extent of damage, ,... are model parameters, For completion; Among them, ElasticNet is introduced to optimize the polynomial regression model by combining L1 regularization and L2 regularization, and its loss function is: In the formula, is the true value, is the predicted value, is the regression coefficient of the jth feature, is the sample size, is the feature vector, is the regularization parameter, which is used to control the strength of regularization. and are the strengths of L1 regularization and L2 regularization, respectively.

9. A flood disaster monitoring and early warning method, characterized in that: The method is implemented according to the urban flood disaster monitoring and early warning system described in claim 1.

Images