Rain type runoff analysis system based on big data mining

Abstract

The invention discloses a rain pattern runoff analysis system based on big data mining, which relates to the technical field of meteorological analysis and comprises a collection and preprocessing module, a rain pattern classification module, a rain pattern clustering module, a runoff generation module, a relation mining module and a real-time monitoring module. According to the rain type runoff analysis system based on big data mining, a complex nonlinear relationship between a rain type and runoff is effectively captured through setting a prediction model with related functions, a deep neural network, a long-short-term memory network and an adaptive algorithm; according to the method, model parameters can be dynamically adjusted to adapt to different geographical and climate conditions, more accurate runoff prediction can be provided in extreme weather events, the defects of a traditional model are overcome, the runoff prediction accuracy under complex terrain conditions is remarkably improved, and more reliable data support is provided for flood and disaster prevention and water resource management.

Description

Technical Field

[0001] The present invention relates to the technical field of meteorological analysis, and in particular to a rain-type runoff analysis system based on big data mining. Background Art

[0002] With the intensification of global climate change, the frequent occurrence of extreme weather events has put forward higher requirements for water resource management and flood prevention and disaster relief work. In the study of hydrological processes caused by rainfall, the relationship between rainfall pattern and runoff has always been the focus of attention in this field. Traditional rainfall-runoff models are usually based on simple physical equations or empirical formulas. These models often assume that the relationship between rainfall pattern and runoff is linear or approximately linear. However, in reality, the relationship between rainfall pattern and runoff is usually complex and nonlinear, and is affected by many factors, such as topography, soil moisture, rainfall intensity and duration, etc. Traditional models have difficulty capturing these complex dynamic interactions, resulting in limited accuracy and reliability of runoff prediction when dealing with complex terrain conditions or extreme weather events.

[0003] Existing rainfall-runoff models mainly rely on linear or simple nonlinear physical or empirical models, which cannot fully capture the complex nonlinear relationship between rainfall patterns and runoff. For example, although the SCS curve number model and the Horton model can provide relatively accurate predictions in some cases, they often appear to be powerless when faced with variable rainfall patterns and complex terrain. In addition, existing clustering and classification algorithms are usually unable to adaptively process large amounts of heterogeneous rainfall data, resulting in the possibility that subtle but important patterns may be ignored during the analysis process. Therefore, traditional models often show problems of inaccurate predictions and poor applicability when dealing with diverse climatic and geographical conditions. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a rain-type runoff analysis system based on big data mining to solve the problems raised in the above background technology.

[0005] To achieve the above-mentioned object, the present invention provides the following technical solutions: a rain pattern runoff analysis system based on big data mining, comprising a collection and preprocessing module, a rain pattern classification module, a rain pattern clustering module, a runoff generation module, a relationship mining module and a real-time monitoring module;

[0006] The collection and preprocessing module collects relevant rainfall pattern and runoff data from multiple data sources, preprocesses the collected data, standardizes the preprocessed data, and integrates data from different data sources;

[0007] The rain type classification module classifies the collected rainfall data based on big data mining technology, and classifies different types of rainfall types according to the characteristics of the rainfall data;

[0008] The rain pattern clustering module groups rain patterns with similar characteristics by clustering technology, analyzes the spatial and temporal distribution of rain patterns, and provides a reference for runoff analysis;

[0009] The runoff generation module uses a physical model to simulate the runoff generation process according to different rainfall patterns and geographical conditions;

[0010] The relationship mining module analyzes the complex nonlinear relationship between rainfall patterns and runoff through big data mining technology, captures the complex dynamic interaction between rainfall patterns and runoff, and provides runoff prediction results;

[0011] The real-time monitoring module is used to obtain the latest meteorological data and update the rain type classification and runoff prediction results in real time when rainfall occurs.

[0012] To further optimize this technical solution, in the collection and preprocessing module:

[0013] The data of rainfall patterns and runoff include meteorological stations, remote sensing data, geographic information system (GIS) data, and historical hydrological data;

[0014] The preprocessing includes denoising, missing value filling, and coordinate system conversion; the preprocessing is based on applying distributed computing technology, such as Hadoop or Spark, to process data.

[0015] To further optimize the technical solution, the rain type classification module has a built-in multi-layer adaptive network classification model (MANCM), which is based on a multi-layer neural network, and its input layer includes multiple meteorological feature vectors, including rainfall intensity (I), duration (T), geographical location (L), and rainfall pattern (P);

[0016] The model presets an adaptive weight parameter (α i ), which is used to adjust the influence of each feature in the classification process according to different regions and climate conditions. This parameter is defined by the formula:

[0017]

[0018] Among them, (β i ) is the basis weight of each feature, (Z) is the state vector containing geographical and climate variables;

[0019] At the same time, the model also optimizes each weight parameter by gradient descent method.

[0020] To further optimize the technical solution, the rain type clustering module has a built-in rain type clustering model, which combines dynamic time warping and density clustering algorithms to capture the spatiotemporal correlation in the rain type data;

[0021] The model first calculates the similarity matrix between each rain type sample, which is defined as:

[0022]

[0023] Among them, (X i ) and (X j ) are two rain type samples, (DTW(X i , X j )) is the dynamic time warping distance between two samples, (σ) is the adjustment parameter used to control the decay rate of similarity;

[0024] The model uses a density clustering algorithm to cluster rain pattern groups with similar spatiotemporal characteristics by defining an adaptive neighborhood parameter (∈i). The specific neighborhood parameter is defined by the following formula:

[0025] ∈ i =γ·median(S ij ), j∈Neighborhood of i

[0026] Among them, (γ) is the clustering sensitivity parameter.

[0027] To further optimize the technical solution, a physical model is built into the runoff generation module, which is a dynamic terrain-rainfall response model (DTRRM), which dynamically adjusts the parameters of the runoff generation process by combining terrain characteristics, soil type and clustered rain patterns;

[0028] The dynamic terrain-rain pattern response model is as follows:

[0029]

[0030] in,

[0031] (Q) is the runoff, i.e. the target output;

[0032] (P) is the actual rainfall, which is a characteristic of rain type;

[0033] (I a ) is the initial loss, i.e. the amount of water absorbed and evaporated by the soil at the initial stage of rainfall;

[0034] (S) is the soil water storage capacity, which is dynamically calculated based on soil type and terrain conditions;

[0035] (λ) is the model adaptive coefficient, which is adjusted according to the terrain slope and soil permeability;

[0036] (m) is the regulation index, which controls the nonlinear relationship between rainfall and runoff and is fitted based on historical data;

[0037] (A) is the drainage area, which is a topographic feature;

[0038] (T) is the rainfall duration, reflecting the temporal distribution of rainfall intensity;

[0039] (ω) is the soil moisture coefficient, which reflects the current soil moisture level;

[0040] (H) is the terrain slope, which affects the speed of rainfall runoff.

[0041] To further optimize the technical solution, the relationship mining module constructs a rainfall pattern-runoff dynamic response network model based on big data mining technology. The rainfall pattern-runoff dynamic response network model extracts features and performs dynamic response based on a deep neural network, specifically including:

[0042] Processing input rain pattern data such as rainfall intensity, rainfall duration, terrain characteristics, soil moisture in a deep neural network;

[0043] The input data is transformed by nonlinear activation functions such as ReLU to extract key patterns that affect runoff generation;

[0044] A long short-term memory network (LSTM) is used to capture long-term dependencies in time series data, memorize the impact of past rainfall patterns on current runoff, and predict future runoff;

[0045] The Bayesian optimization algorithm is introduced to automatically adjust the model's hyperparameters, such as the learning rate, the number of hidden layer units, and the regularization coefficient. The optimization objective function is as follows:

[0046]

[0047] in, is the actual runoff, (Q i ) is the predicted value, and (λ) is the regularization coefficient used to prevent overfitting.

[0048] To further optimize the technical solution, the real-time monitoring module is combined with a geographic information system (GIS) and Internet of Things (IoT) devices, such as rain gauges and water level gauges, to monitor hydrological conditions in real time and predict runoff based on real-time rainfall patterns. If potential flood risks are detected, the module will automatically generate early warning information and notify through SMS, email or social media channels.

[0049] To further optimize the technical solution, the system also includes a data analysis module to analyze historical rainfall and runoff data to discover long-term trends and anomalies.

[0050] Further optimizing the technical solution, the data analysis module automatically detects outliers in historical data, such as extreme rainfall events or rare runoff patterns, based on big data mining technology;

[0051] The module also supports users to conduct hypothesis testing to analyze the changes in the frequency of a particular type of rainfall pattern in different years in the past.

[0052] To further optimize the technical solution, the system also includes a data visualization module, which dynamically generates various visualization charts, such as heat maps, time series graphs, and spatial distribution graphs based on the data and results generated by the system, to display the temporal evolution and spatial distribution characteristics of rainfall patterns, as well as the changes in runoff under different rainfall patterns.

[0053] Compared with the prior art, the present invention provides a rain-type runoff analysis system based on big data mining, which has the following beneficial effects:

[0054] This rainfall pattern runoff analysis system based on big data mining, by setting up a prediction model with relevant functions, effectively captures the complex nonlinear relationship between rainfall pattern and runoff through deep neural networks, long short-term memory networks and adaptive algorithms; it can not only dynamically adjust model parameters to adapt to different geographical and climatic conditions, but also provide more accurate runoff predictions in extreme weather events, filling the shortcomings of traditional models, significantly improving the accuracy of runoff predictions under complex terrain conditions, and providing more reliable data support for flood control and water resources management. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a structural schematic diagram of a rain-type runoff analysis system based on big data mining proposed by the present invention;

[0056] Figure 2 A flow chart of a data analysis module in a rain-type runoff analysis system based on big data mining proposed by the present invention;

[0057] Figure 3 This is a chart type diagram of a visualization module in a rain-type runoff analysis system based on big data mining proposed by the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] Embodiment 1:

[0060] See also Figure 1 ,A rain type runoff analysis system based on big data mining includes a collection and preprocessing module, a rain type classification module, a rain type clustering module, a runoff generation module, a relationship mining module and a real-time monitoring module.

[0061] The collection and preprocessing module collects relevant rainfall pattern and runoff data from multiple data sources, preprocesses the collected data, standardizes the preprocessed data, integrates data from different data sources, ensures that data from different sources can be seamlessly integrated, and provides a unified basis for subsequent analysis.

[0062] In this embodiment, in the collection and preprocessing module:

[0063] The data of rainfall patterns and runoff include meteorological stations, remote sensing data, geographic information system (GIS) data, and historical hydrological data;

[0064] The preprocessing includes denoising, missing value filling, and coordinate system conversion; the preprocessing is based on applying distributed computing technology, such as Hadoop or Spark, to process data.

[0065] The rain type classification module classifies the collected rainfall data based on big data mining technology, and classifies different types of rainfall types according to the characteristics of the rainfall data.

[0066] In this embodiment, the rain type classification module has a built-in multi-layer adaptive network classification model (MANCM), which is based on a multi-layer neural network, and its input layer includes multiple meteorological feature vectors, including rainfall intensity (I), duration (T), geographical location (L), and rainfall pattern (P);

[0067] The model presets an adaptive weight parameter (α i) to adjust the influence of each feature in the classification process according to different regions and climate conditions. The parameter is defined by the formula:

[0068]

[0069] Among them, (β i ) is the basis weight of each feature, (Z) is the state vector containing geographical and climate variables;

[0070] At the same time, the model also optimizes each weight parameter through the gradient descent method, so that the classification results can maintain high accuracy in different regions and climatic conditions.

[0071] In actual application, the MANCM model receives pre-processed rainfall data as input and classifies various types of rainfall by adaptively adjusting the weight of each feature. The classification results include specific types such as showers, continuous rainfall, and thunderstorms. The system compares the classification results with historical data and automatically identifies abnormal situations, such as extreme weather events. In addition, the classification results also provide key input data for subsequent runoff analysis.

[0072] The rain pattern clustering module groups rain patterns with similar characteristics by clustering technology, analyzes the spatial and temporal distribution of rain patterns, and provides a reference for runoff analysis.

[0073] In this embodiment, the rain pattern clustering module has a built-in rain pattern clustering model, which combines dynamic time warping and density clustering algorithms to capture the spatiotemporal correlation in the rain pattern data;

[0074] The model first calculates the similarity matrix between each rain type sample, which is defined as:

[0075]

[0076] Among them, (X i ) and (X j ) are two rain type samples, (DTW(X i , X j )) is the dynamic time warping distance between two samples, (σ) is the adjustment parameter used to control the decay rate of similarity;

[0077] The model uses a density clustering algorithm to cluster rain pattern groups with similar spatiotemporal characteristics by defining an adaptive neighborhood parameter (∈i). The specific neighborhood parameter is defined by the following formula:

[0078] ∈ i =γ·median(S ij ), j∈Neighborhood of i

[0079] Among them, (γ) is the clustering sensitivity parameter.

[0080] In cluster analysis, the model can identify rainfall events that are close in space and continuous in time. For example, the system can reveal the rainfall types and their evolution patterns that occur frequently in a specific area in a specific season through cluster analysis. This information provides valuable reference data for subsequent runoff prediction. In addition, the system can also monitor changes in rainfall patterns through dynamic clustering results, detect abnormal phenomena in a timely manner, and warn of potential hydrological risks.

[0081] The runoff generation module uses a physical model to simulate the runoff generation process according to different rainfall patterns and geographical conditions.

[0082] In this embodiment, the runoff generation module has a built-in physical model, which is a dynamic terrain-rain pattern response model (DTRRM), which dynamically adjusts the parameters of the runoff generation process by combining terrain characteristics, soil type and clustered rain pattern;

[0083] The dynamic terrain-rain pattern response model is as follows:

[0084]

[0085] in

[0086] (Q) is the runoff, i.e. the target output;

[0087] (P) is the actual rainfall, which is a characteristic of rain type;

[0088] (I a ) is the initial loss, i.e. the amount of water absorbed and evaporated by the soil at the initial stage of rainfall;

[0089] (S) is the soil water storage capacity, which is dynamically calculated based on soil type and terrain conditions;

[0090] (λ) is the model adaptive coefficient, which is adjusted according to the terrain slope and soil permeability;

[0091] (m) is the regulation index, which controls the nonlinear relationship between rainfall and runoff and is fitted based on historical data;

[0092] (A) is the drainage area, which is a topographic feature;

[0093] (T) is the rainfall duration, reflecting the temporal distribution of rainfall intensity;

[0094] (ω) is the soil moisture coefficient, which reflects the current soil moisture level;

[0095] (H) is the terrain slope, which affects the speed of rainfall runoff.

[0096] Several core parameters in the model, such as (λ) and (m), are dynamically adjusted and optimized. In particular, the model introduces an exponential term Used to describe the complex effects of rainfall duration and soil moisture on runoff generation. This term reflects that runoff generation will be faster under conditions of higher slopes or higher soil moisture.

[0097] In practical applications, the DTRRM model receives input data such as rainfall, soil water storage capacity, and terrain slope. The model first calculates the initial loss (I a ) and soil water storage capacity (S), and then dynamically adjust the adaptive coefficient (λ) and regulation index (m) to adapt to different terrain conditions and rainfall patterns.

[0098] By inputting rainfall duration (T) and current soil moisture coefficient (ω), the model calculates the final runoff volume (Q) and passes it as output to subsequent modules.

[0099] The innovation of the DTRRM model is that it considers the nonlinear combination of multiple influencing factors, which enables the model to more accurately simulate the runoff generation process under complex terrain and variable rainfall patterns. The adaptability of the model enables it to automatically adjust key parameters to cope with different geographical and climatic conditions, thereby improving the accuracy and reliability of runoff prediction.

[0100] The relationship mining module analyzes the complex nonlinear relationship between rainfall patterns and runoff through big data mining technology, captures the complex dynamic interactions between rainfall patterns and runoff, and provides runoff prediction results.

[0101] In this embodiment, the relationship mining module constructs a rainfall pattern-runoff dynamic response network model based on big data mining technology. The rainfall pattern-runoff dynamic response network model extracts features and performs dynamic response based on a deep neural network, specifically including:

[0102] Processing input rain pattern data such as rainfall intensity, rainfall duration, terrain characteristics, soil moisture in a deep neural network;

[0103] The input data is transformed by nonlinear activation functions such as ReLU to extract key patterns that affect runoff generation;

[0104] A long short-term memory network (LSTM) is used to capture long-term dependencies in time series data, memorize the impact of past rainfall patterns on current runoff, and predict future runoff;

[0105] The Bayesian optimization algorithm is introduced to automatically adjust the model's hyperparameters, such as the learning rate, the number of hidden layer units, and the regularization coefficient. The optimization objective function is as follows:

[0106]

[0107] in, is the actual runoff, (Q i ) is the predicted value, and (λ) is the regularization coefficient used to prevent overfitting.

[0108] In actual application, the model receives rainfall data as input, performs layer-by-layer calculations through feature extraction and dynamic response, and finally outputs the runoff prediction value for each time step. Through the Bayesian optimization algorithm, the system can continuously adjust the model parameters so that the prediction results remain highly accurate under different rainfall patterns and geographical conditions.

[0109] The model's dynamic response is particularly good at handling long-term rainfall events, such as continuous rainfall or extreme weather conditions. Through the LSTM network, it can effectively remember and use past rainfall information to predict future runoff changes. This ability enables the model to provide more accurate runoff prediction results when dealing with complex hydrological conditions, especially in extreme weather events.

[0110] The real-time monitoring module is used to obtain the latest meteorological data and update the rain type classification and runoff prediction results in real time when rainfall occurs.

[0111] In this embodiment, the real-time monitoring module combines geographic information system (GIS) and Internet of Things (IoT) devices, such as rain gauges and water level gauges, to monitor hydrological conditions in real time and predict runoff based on real-time rainfall patterns. If potential flood risks are detected, the module will automatically generate warning information and notify through SMS, email or social media channels.

[0112] Embodiment 2:

[0113] See also Figure 2 and Figure 3 Based on the rain-type runoff analysis system based on big data mining described in Example 1, the system also includes a data analysis module to analyze historical rainfall and runoff data to discover long-term trends and anomalies. Based on big data mining technology, outliers in historical data, such as extreme rainfall events or rare runoff patterns, are automatically detected. This not only helps to understand past hydrological events, but also provides a reference for future water resources management.

[0114] The module also supports users to conduct hypothesis testing to analyze the changes in the frequency of a particular type of rainfall pattern in different years in the past.

[0115] In this embodiment, the system also includes a data visualization module, which dynamically generates various visualization charts, such as heat maps, time series charts, and spatial distribution charts, based on the data and results generated by the system, to display the temporal evolution and spatial distribution characteristics of rainfall patterns, as well as the runoff changes under different rainfall patterns. Users can customize the viewing angle and display content through operations such as dragging and zooming, and further view the in-depth information of the data.

[0116] The beneficial effects of the present invention are:

[0117] This rainfall pattern runoff analysis system based on big data mining, by setting up a prediction model with relevant functions, effectively captures the complex nonlinear relationship between rainfall pattern and runoff through deep neural networks, long short-term memory networks and adaptive algorithms; it can not only dynamically adjust model parameters to adapt to different geographical and climatic conditions, but also provide more accurate runoff predictions in extreme weather events, filling the shortcomings of traditional models, significantly improving the accuracy of runoff predictions under complex terrain conditions, and providing more reliable data support for flood control and water resources management.

[0118] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0119] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A rain-type runoff analysis system based on big data mining, characterized in that: It includes collection and preprocessing module, rain type classification module, rain type clustering module, runoff generation module, relationship mining module and real-time monitoring module; The collection and preprocessing module collects relevant rainfall pattern and runoff data from multiple data sources, preprocesses the collected data, standardizes the preprocessed data, and integrates data from different data sources; The rain type classification module classifies the collected rainfall data based on big data mining technology, and classifies different types of rainfall types according to the characteristics of the rainfall data; The rain pattern clustering module groups rain patterns with similar characteristics by clustering technology, analyzes the spatial and temporal distribution of rain patterns, and provides a reference for runoff analysis; The runoff generation module uses a physical model to simulate the runoff generation process according to different rainfall patterns and geographical conditions; The relationship mining module analyzes the complex nonlinear relationship between rainfall patterns and runoff through big data mining technology, captures the complex dynamic interaction between rainfall patterns and runoff, and provides runoff prediction results; The real-time monitoring module is used to obtain the latest meteorological data and update the rain type classification and runoff prediction results in real time when rainfall occurs.

2. The rain-type runoff analysis system based on big data mining according to claim 1 is characterized in that: In the collection and preprocessing module: The data of rainfall patterns and runoff include meteorological stations, remote sensing data, geographic information system (GIS) data, and historical hydrological data; The preprocessing includes denoising, missing value filling, and coordinate system conversion; the preprocessing is based on applying distributed computing technology, such as Hadoop or Spark, to process data.

3. The rain-type runoff analysis system based on big data mining according to claim 1 is characterized in that: The rain type classification module has a built-in multi-layer adaptive network classification model (MANCM), which is based on a multi-layer neural network and whose input layer includes multiple meteorological feature vectors, including rainfall intensity (I), duration (T), geographical location (L), and rainfall pattern (P); The model presets an adaptive weight parameter (α i ), which is used to adjust the influence of each feature in the classification process according to different regions and climate conditions. This parameter is defined by the formula: Among them, (β i ) is the basis weight of each feature, (Z) is the state vector containing geographical and climate variables; At the same time, the model also optimizes each weight parameter by gradient descent method.

4. The rain-type runoff analysis system based on big data mining according to claim 1 is characterized in that: The rain pattern clustering module has a built-in rain pattern clustering model, which combines dynamic time warping and density clustering algorithms to capture the spatiotemporal correlation in rain pattern data; The model first calculates the similarity matrix between each rain type sample, which is defined as: Among them, (X i ) and (X j ) are two rain type samples, (DTW(X i , X j )) is the dynamic time warping distance between two samples, (σ) is the adjustment parameter used to control the decay rate of similarity; The model uses a density clustering algorithm to cluster rain pattern groups with similar spatiotemporal characteristics by defining an adaptive neighborhood parameter (∈i). The specific neighborhood parameter is defined by the following formula: ∈ i =γ·median(S ij ),j∈Neighborhood of i Among them, (γ) is the clustering sensitivity parameter.

5. The rain-type runoff analysis system based on big data mining according to claim 1 is characterized in that: The runoff generation module has a built-in physical model, which is a dynamic terrain-rain pattern response model (DTRRM), which dynamically adjusts the parameters of the runoff generation process by combining terrain characteristics, soil types and clustered rain patterns; The dynamic terrain-rain pattern response model is as follows: in, (Q) is the runoff, i.e. the target output; (P) is the actual rainfall, which is a characteristic of rain type; (I a ) is the initial loss, i.e. the amount of water absorbed and evaporated by the soil at the initial stage of rainfall; (S) is the soil water storage capacity, which is dynamically calculated based on soil type and terrain conditions; (λ) is the model adaptive coefficient, which is adjusted according to the terrain slope and soil permeability; (m) is the regulation index, which controls the nonlinear relationship between rainfall and runoff and is fitted based on historical data; (A) is the drainage area, which is a topographic feature; (T) is the rainfall duration, reflecting the temporal distribution of rainfall intensity; (ω) is the soil moisture coefficient, which reflects the current soil moisture level; (H) is the terrain slope, which affects the speed of rainfall runoff.

6. The rain-type runoff analysis system based on big data mining according to claim 1 is characterized in that: The relationship mining module constructs a rainfall pattern-runoff dynamic response network model based on big data mining technology. The rainfall pattern-runoff dynamic response network model extracts features and performs dynamic response based on a deep neural network, specifically including: Processing input rain pattern data such as rainfall intensity, rainfall duration, terrain characteristics, soil moisture in a deep neural network; The input data is transformed by nonlinear activation functions such as ReLU to extract key patterns that affect runoff generation; Long short-term memory (LSTM) networks are used to capture long-term dependencies in time series data, memorize the impact of past rainfall patterns on current runoff, and predict future runoff. The Bayesian optimization algorithm is introduced to automatically adjust the model's hyperparameters, such as the learning rate, the number of hidden layer units, and the regularization coefficient. The optimization objective function is as follows: in, is the actual runoff, (Q i ) is the predicted value, and (λ) is the regularization coefficient used to prevent overfitting.

7. The rain-type runoff analysis system based on big data mining according to claim 1 is characterized in that: The real-time monitoring module combines geographic information systems (GIS) and Internet of Things (IoT) devices, such as rain gauges and water level gauges, to monitor hydrological conditions in real time and predict runoff based on real-time rainfall patterns. If potential flood risks are detected, the module will automatically generate warning information and notify through SMS, email or social media channels.

8. The rain-type runoff analysis system based on big data mining according to claim 1 is characterized in that: The system also includes a data analytics module that analyzes historical rainfall and runoff data to identify long-term trends and anomalies.

9. The rain-type runoff analysis system based on big data mining according to claim 8 is characterized in that: The data analysis module automatically detects outliers in historical data, such as extreme rainfall events or rare runoff patterns, based on big data mining technology; The module also supports users to conduct hypothesis testing to analyze the changes in the frequency of a particular type of rainfall pattern in different years in the past.

10. The rain-type runoff analysis system based on big data mining according to claim 1, characterized in that: The system also includes a data visualization module, which dynamically generates various visualization charts, such as heat maps, time series graphs, and spatial distribution graphs based on the data and results generated by the system, showing the temporal evolution and spatial distribution characteristics of rainfall patterns, as well as the changes in runoff under different rainfall patterns.

Images