Typhoon rainfall prediction method and system based on machine learning and BMA model

Abstract

The embodiment of the invention provides a typhoon rainfall prediction method and system based on machine learning and a BMA model, and belongs to the technical field of extreme weather event prediction. According to the invention, feature information can be obtained; the feature information comprises typhoon features, grid space attributes and meteorological features; acquiring a machine learning model according to the feature information; the machine learning model comprises an RF model, an SVM model, a KNN model and an XGBoost model; obtaining a BMA model according to the machine learning model; and obtaining typhoon rainfall prediction data according to the BMA model. The accuracy and interpretation of typhoon rainfall prediction can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of extreme weather event prediction, and in particular to a typhoon rainfall prediction method and system based on machine learning and BMA model. Background Art

[0002] The existing technologies for typhoon rainfall forecasting currently have the following main defects:

[0003] Existing research primarily focuses on station-scale rainfall forecasts, but fails to effectively account for the spatial heterogeneity within a typhoon's impact zone. Precipitation distribution within a typhoon's impact zone exhibits significant spatial variation. Therefore, accurately predicting rainfall distribution across the entire typhoon-affected region is often difficult if forecasts based solely on data from a single station. This means that many local characteristics of typhoon rainfall may be overlooked, resulting in forecasts that fail to fully reflect the typhoon's actual impact.

[0004] Many current typhoon rainfall forecasting methods rely on selecting a single optimal model. However, this optimal model selection often relies on a specific dataset and model parameter settings, making the forecast results susceptible to data quality, model assumptions, or parameter settings, leading to high forecast uncertainty. Furthermore, ignoring the uncertainty in the model selection process can also lead to a lack of robustness in the forecast results, making them inaccurate and inefficient across different scenarios or conditions. Summary of the Invention

[0005] The main purpose of the embodiments of the present application is to provide a typhoon rainfall prediction method and system based on machine learning and BMA model.

[0006] The technical solution adopted by the present invention is:

[0007] In one aspect, an embodiment of the present invention provides a typhoon rainfall prediction method based on machine learning and a BMA model, the typhoon rainfall prediction method based on machine learning and a BMA model comprising the following steps:

[0008] Acquiring characteristic information; the characteristic information includes typhoon characteristics, grid space attributes and meteorological characteristics;

[0009] According to the feature information, a machine learning model is obtained; the machine learning model includes a RF model, a SVM model, a KNN model and an XGBoost model;

[0010] According to the machine learning model, a BMA model is obtained;

[0011] Typhoon rainfall forecast data is obtained according to the BMA model.

[0012] Furthermore, the obtaining of characteristic information includes the following steps:

[0013] Obtain typhoon center data;

[0014] constructing a buffer zone based on the typhoon center data by using a grid data extraction method;

[0015] Obtaining characteristic information according to the buffer zone; the characteristic information includes typhoon characteristics, grid space attributes and meteorological characteristics;

[0016] The typhoon characteristics include the longitude of the typhoon center, the latitude of the typhoon center, the maximum sustained wind speed, the weighted typhoon wind speed, the typhoon center air pressure, the weighted typhoon center air pressure, the moving speed of the typhoon center, the weighted typhoon moving speed, and the moving direction of the typhoon center; the weighted typhoon wind speed, the weighted typhoon center air pressure, and the weighted typhoon moving speed are the dynamic weighted typhoon characteristics in the typhoon characteristics; the maximum sustained wind speed, the typhoon center air pressure, and the typhoon center moving speed are the original typhoon characteristics of the typhoon center in the typhoon characteristics;

[0017] The grid spatial attributes include the grid latitude, grid longitude, elevation information, the distance of the grid from the typhoon center, and the orientation of the grid relative to the typhoon center of each grid;

[0018] The meteorological characteristics include rainfall, temperature, relative humidity, wind speed and wind direction of each grid;

[0019] The lag time of the rainfall is selected by the autocorrelation coefficient;

[0020] The formula used to calculate the dynamic weighted typhoon characteristics in each grid includes:

[0021] ω i =1-d i / r;

[0022] Weighted_f i =f·ω i ;

[0023] Where i is a natural number: d i is the distance between the ith grid and the typhoon center; r is the radius of the buffer zone; ω i is the dynamic weight of the i-th grid; f represents the original typhoon characteristics; Weighted_f i is the dynamic weighted typhoon characteristic of the i-th grid.

[0024] Furthermore, the lag time of the rainfall is selected by the autocorrelation coefficient, and the formula used includes:

[0025]

[0026] Where N is the total number of observations in the time series; t is the first moment, tk is the second moment; R t,i is the rainfall of the grid at the first moment; R t-k,i is the rainfall of the grid at the second moment; is the average rainfall of the grid at the first moment; is the average rainfall of the grid at the second moment; k is an integer, 1≤k≤24; ρ(k) is the lag time of rainfall.

[0027] Furthermore, obtaining a machine learning model based on the feature information includes the following steps:

[0028] The feature information is processed using a Z-score normalization method, and each feature information is rescaled so that all the feature information has a zero mean and a unit standard deviation, thereby obtaining model training data;

[0029] Training an initial training model according to the model training data; the initial training model includes an initial RF model, an initial SVM model, an initial KNN model and an initial XGBoost model;

[0030] Evaluate the performance of each of the initial training models through five-fold cross validation, and use a grid search algorithm to perform hyperparameter tuning to obtain the optimal hyperparameter combination for each of the initial training models;

[0031] According to the optimal hyperparameter combination, a machine learning model is obtained; the machine learning model includes an RF model, an SVM model, a KNN model and an XGBoost model.

[0032] Furthermore, obtaining the BMA model according to the machine learning model includes the following steps:

[0033] Obtaining prediction results of each of the machine learning models;

[0034] Obtaining a model weight according to the prediction result;

[0035] According to the model weights, a BMA model is obtained.

[0036] Furthermore, the model weight is obtained according to the prediction result, and the formula used includes:

[0037]

[0038] Among them, M j is the model prediction value of the j-th machine learning model; D = [y1, y2, ..., y N ] represents the observed data; P(D|Mj ) is the predicted value M of the given model j The conditional probability of rainfall data; P(M j ) is the prior probability of the j-th machine learning model; M i is the model prediction value of the i-th machine learning model; P(M i ) is the prior probability of the i-th machine learning model; P(M i |D) is the model prediction value M i The posterior probability of P(M j |D) is the model prediction value M j The posterior probability, w j is the model weight, satisfying

[0039] Furthermore, the typhoon rainfall prediction method based on machine learning and the BMA model further includes the following steps:

[0040] Get rainfall observations;

[0041] Get the predicted value of the BMA model;

[0042] Evaluation data is obtained based on the rainfall observation value and the predicted value; the evaluation data includes a determination coefficient, a mean absolute error, and a root mean square error.

[0043] Furthermore, the evaluation data is obtained based on the rainfall observation value and the predicted value, and the formula used includes:

[0044]

[0045] in, is the i-th rainfall observation value; is the i-th predicted value; is the average value of the rainfall observations; R 2 is the coefficient of determination; MAE is the mean absolute error; RMSE is the root mean square error.

[0046] On the other hand, an embodiment of the present invention also provides a typhoon rainfall prediction system based on machine learning and BMA model, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, it implements the typhoon rainfall prediction method based on machine learning and BMA model as described above.

[0047] On the other hand, an embodiment of the present invention also provides a computer-readable storage medium, which stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the typhoon rainfall prediction method based on machine learning and BMA model as described above.

[0048] The embodiments of this application include at least the following beneficial effects: This application provides a typhoon rainfall prediction method and system based on machine learning and the BMA model. The present invention can obtain feature information; the feature information includes typhoon characteristics, grid spatial attributes, and meteorological characteristics; obtain a machine learning model based on the feature information; the machine learning model includes a RF model, a SVM model, a KNN model, and an XGBoost model; obtain a BMA model based on the machine learning model; and obtain typhoon rainfall forecast data based on the BMA model. This invention can improve the accuracy and explanatory power of typhoon rainfall forecasts. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A typhoon rainfall prediction method based on machine learning and the BMA model is provided in an embodiment of the present invention;

[0050] Figure 2 is a schematic diagram of the study area and grid extraction provided by an embodiment of the present invention;

[0051] Figure 3 These are 28 typhoon paths and 462 typhoon centers (2020-2023; 3-hour intervals) provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present application. They are merely examples of devices and methods consistent with some aspects of the embodiments of the present application as detailed in the appended claims.

[0053] It will be understood that the terms "first", "second", etc. used in this application may be used herein to describe various concepts, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" and "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0054] The terms "at least one", "plurality", "each", "any", etc. used in this application include "at least one", "two" or more, "plurality" or "each", "any" or "any one", "each" or "any one" as used herein.

[0055] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0056] Before explaining the embodiments of the present application in detail, some of the nouns and terms involved in the embodiments of the present application are first explained. The nouns and terms involved in the embodiments of the present application are subject to the following explanations.

[0057] 1) Bayesian Model Averaging (BMA) is an ensemble learning method based on Bayesian theory that obtains the final prediction by weighted averaging the prediction results of multiple models.

[0058] 2) RF model (Random Forest), Random Forest (RF) is an ensemble learning method and an extension of the decision tree;

[0059] 3) SVM model (Support Vector Machine), Support Vector Machine (SVM), is a supervised learning model;

[0060] 4) KNN model (K-Nearest Neighbors), K-nearest neighbors (KNN), is an instance-based learning algorithm;

[0061] 5) XGBoost (Extreme Gradient Boosting) model. XGBoost is an efficient gradient boosting tree (GBDT) algorithm.

[0062] 6) Coefficient of determination (R 2 ), which is an evaluation indicator in regression analysis and is used to measure how well the model fits the data;

[0063] 7) Mean Absolute Error (MAE) is an evaluation metric for regression models, representing the average absolute difference between the predicted value and the true value;

[0064] 8) Root Mean Squared Error (RMSE) is an evaluation metric for regression models, representing the average of the square roots of the differences between the predicted values ​​and the true values.

[0065] 9) Z-score normalization is a data preprocessing method;

[0066] 10) 5-fold cross-validation is a model evaluation method;

[0067] 11) Grid Search algorithm is a method for hyperparameter optimization.

[0068] The embodiments of the present invention are further described below with reference to the accompanying drawings.

[0069] On the one hand, the embodiment of the present invention provides a typhoon rainfall prediction method based on machine learning and BMA model, referring to Figure 1 ,The typhoon rainfall prediction method based on machine learning and BMA model,includes the following steps:

[0070] S100, acquiring characteristic information; the characteristic information includes typhoon characteristics, grid space attributes, and meteorological characteristics;

[0071] S200, obtaining a machine learning model based on the feature information; the machine learning model includes a RF model, a SVM model, a KNN model, and an XGBoost model;

[0072] S300, obtaining a BMA model according to the machine learning model;

[0073] S400. Obtain typhoon rainfall forecast data based on the BMA model.

[0074] Step S100 disclosed in the embodiment of the present invention obtains characteristic information, including the following steps:

[0075] S110, obtain typhoon center data;

[0076] S120, constructing a buffer zone using a grid data extraction method based on typhoon center data;

[0077] S130, obtaining characteristic information according to the buffer zone; the characteristic information includes typhoon characteristics, grid space attributes, and meteorological characteristics;

[0078] S140. Typhoon characteristics include the longitude of the typhoon center, the latitude of the typhoon center, the maximum sustained wind speed, the weighted typhoon wind speed, the typhoon center air pressure, the weighted typhoon center air pressure, the typhoon center moving speed, the weighted typhoon moving speed, and the moving direction of the typhoon center. The weighted typhoon wind speed, the weighted typhoon center air pressure, and the weighted typhoon moving speed are the dynamic weighted typhoon characteristics in the typhoon characteristics. The maximum sustained wind speed, the typhoon center air pressure, and the typhoon center moving speed are the original typhoon characteristics of the typhoon center in the typhoon characteristics.

[0079] S150, grid spatial attributes include grid latitude, grid longitude, elevation information, distance from the grid to the typhoon center, and orientation of the grid relative to the typhoon center of each grid;

[0080] S160, meteorological characteristics include rainfall, temperature, relative humidity, wind speed, and wind direction for each grid;

[0081] S170, the lag time of rainfall is selected by the autocorrelation coefficient;

[0082] S180: The formulas used to calculate the dynamic weighted typhoon characteristics in each grid include:

[0083] ω i =1-d i / r;

[0084] Weighted_f i =f·ω i ;

[0085] Where i is a natural number: d i is the distance between the ith grid and the typhoon center; r is the radius of the buffer zone; ω i is the dynamic weight of the i-th grid; f represents the original typhoon characteristics; Weighted_f i is the dynamic weighted typhoon characteristic of the i-th grid.

[0086] In step S170 disclosed in the embodiment of the present invention, the lag time of rainfall is selected by the autocorrelation coefficient, and the formula used includes:

[0087]

[0088] Where N is the total number of observations in the time series; t is the first moment, tk is the second moment; R t,i is the rainfall of the grid at the first moment; R t-k,i is the rainfall of the grid at the second moment; is the average rainfall of the grid at the first moment; is the average rainfall of the grid at the second moment; k is an integer, 1≤k≤24; ρ(k) is the lag time of rainfall.

[0089] Step S200 disclosed in the embodiment of the present invention obtains a machine learning model based on feature information, including the following steps:

[0090] S210, processing the feature information using a Z-score standardization method, rescaling each feature information so that all feature information has zero mean and unit standard deviation, thereby obtaining model training data;

[0091] S220, training an initial training model according to the model training data; the initial training model includes an initial RF model, an initial SVM model, an initial KNN model, and an initial XGBoost model;

[0092] S230, evaluating the performance of each initial training model through five-fold cross validation, and performing hyperparameter tuning using a grid search algorithm to obtain the optimal hyperparameter combination for each initial training model;

[0093] S240. Obtain a machine learning model based on the optimal hyperparameter combination; the machine learning model includes a RF model, a SVM model, a KNN model, and an XGBoost model.

[0094] Step S300 disclosed in the embodiment of the present invention obtains a BMA model based on a machine learning model, including the following steps:

[0095] S310, obtaining prediction results of each machine learning model;

[0096] S320, obtaining a model weight according to the prediction result;

[0097] S330: Obtain a BMA model based on the model weights.

[0098] As an optional implementation, Bayesian Model Averaging (BMA) in this embodiment of the present invention is an ensemble learning method used to combine the predictions of multiple models through weighted averaging. The BMA model not only combines the predictions of different models but also quantifies the credibility of each model using Bayesian methods.

[0099] The steps include:

[0100] 1. Model evaluation:

[0101] The prediction results of each machine learning model are evaluated, and indicators such as cross-validation, AUC, and mean square error (MSE) are usually used to judge the prediction ability of each model.

[0102] Based on the evaluation results, the posterior probability of each model is calculated, that is, the weight of each model.

[0103] 2. Model weighting:

[0104] The output of each machine learning model (e.g., precipitation prediction) is weighted and averaged according to its posterior probability.

[0105] The weighting process adjusts the weight of each model based on its credibility, with models with higher accuracy receiving higher weights.

[0106] 3. Bayesian Inference:

[0107] Bayesian inference is used to calculate the posterior probability of each model and use this probability to adjust the model output. The Bayesian inference process effectively integrates the prediction results of each model, ultimately obtaining a weighted typhoon rainfall forecast result.

[0108] In step S320 disclosed in the embodiment of the present invention, the model weight is obtained based on the prediction result, and the formula used includes:

[0109]

[0110] Among them, M j is the model prediction value of the jth machine learning model; D = [y1,y2,...,y N ] represents the observed data; P(D|M j ) is the predicted value M of the given model j The conditional probability of rainfall data; P(M j ) is the prior probability of the j-th machine learning model; M i is the model prediction value of the i-th machine learning model; P(M i ) is the prior probability of the i-th machine learning model; P(M i |D) is the model prediction value M i The posterior probability of P(M j |D) is the model prediction value M j The posterior probability, w j is the model weight, satisfying

[0111] The typhoon rainfall prediction method based on machine learning and the BMA model disclosed in an embodiment of the present invention further includes the following steps:

[0112] S500, obtaining rainfall observation values;

[0113] S600, obtaining the predicted value of the BMA model;

[0114] S700: Obtain evaluation data based on the rainfall observation value and the predicted value; the evaluation data includes the coefficient of determination, the mean absolute error, and the root mean square error.

[0115] Step S700 disclosed in the embodiment of the present invention obtains evaluation data based on the rainfall observation value and the predicted value, and the formula used includes:

[0116]

[0117] in, is the i-th rainfall observation value; is the i-th predicted value; is the average value of rainfall observations; R 2is the coefficient of determination; MAE is the mean absolute error; RMSE is the root mean square error.

[0118] As an optional implementation, the present invention develops a robust BMA model to quantitatively predict hourly typhoon rainfall in the 0-6 hour forecast period in the typhoon-affected area. The specific steps are as follows:

[0119] Step 1: Selection of characteristic variables (characteristic information). First, a buffer zone with a radius of 400 kilometers is constructed based on the typhoon center as the typhoon's impact range. Then, three types of data (characteristic information) are collected within this range: typhoon characteristics, grid spatial attributes, and meteorological characteristics, as shown in Table 1. Typhoon characteristics include parameters such as the location of the typhoon center (including the longitude and latitude of the typhoon center), wind speed (including the maximum sustained wind speed), air pressure (including the air pressure at the typhoon center), the moving speed of the typhoon center, and the moving direction of the typhoon center. In addition, the dynamic weighted typhoon characteristics of each grid (including weighted typhoon wind speed, weighted typhoon center air pressure, and weighted typhoon moving speed) need to be calculated based on the distance between each grid and the typhoon center. The specific calculation method is shown in Equations (1) and (2). Grid spatial attributes include the location of each grid (including grid latitude and grid longitude), elevation information, the distance between the grid and the typhoon center, and the orientation of the grid relative to the typhoon center. Meteorological features include variables such as rainfall, temperature, relative humidity, wind speed, and wind direction for each grid. The lag time of rainfall is selected using the autocorrelation coefficient, as shown in Equation (3), with a threshold set to 0.5. All feature variables are processed using the Z-score standardization method, rescaling each variable to have zero mean and unit standard deviation, as shown in Equation (4). Through standardization, variables of different dimensions can be compared under the same standard, avoiding the impact of data scale on model training.

[0120] ω i =1-d i / r (1)

[0121] Weighted_f i =f·ω i (2)

[0122]

[0123]

[0124] Where: d i is the distance from grid i to the typhoon center (km); r is the buffer zone radius (400km); ω iis the dynamic weight of grid i; f represents the original typhoon characteristics of the typhoon center, including the maximum sustained wind speed (Typhoon_wind), typhoon center pressure (Typhoon_pressure) and typhoon center translation speed (Translation_speed) in Table 1; Weighted_f i is the dynamic weighted typhoon characteristic of grid i; N is the total number of all observations in the time series; k is the lag time, ranging from k = 1, 2, ..., 24; R t,i and R t-k,i are the rainfall amounts of grid i at time t and tk respectively; and is the average rainfall at the corresponding time; X is the original data (feature information); μ is the mean of the variable; σ is the standard deviation of the variable.

[0125] Table 1. Model input

[0126]

[0127] Variable Description:

[0128] Typhoon_lon The longitude of the typhoon's center

[0129] Typhoon_lat The latitude of the typhoon's center

[0130] Typhoon_wind Maximum sustained wind speed

[0131] Weighted_typhoon_wind Weighted typhoon wind speed

[0132] Typhoon_pressure Typhoon center pressure

[0133] Weighted_typhoon_pressure Weighted typhoon center pressure

[0134] Translation_speed The moving speed of the typhoon center

[0135] Weighted_translation_speed Weighted typhoon movement speed

[0136] Translation_direction The direction of movement of the typhoon center

[0137] elevation grid elevation

[0138] distance The distance between the grid and the center of the typhoon

[0139] grid_direction The direction of the grid relative to the typhoon center

[0140] gird_lat grid latitude

[0141] grid_lon grid longitude

[0142] R t-4 ,R t-3 ,R t-2 ,R t-1 Rainfall variables with a lag of t-4 to t-1

[0143] d2m 2m dew point temperature

[0144] t2m 2m temperature

[0145] skt surface temperature

[0146] msl mean sea level pressure

[0147] sp surface pressure

[0148] cbh cloud base height

[0149] rh relative humidity

[0150] u100 u component of wind speed at 100 m

[0151] v100 v component of wind speed at 100 m

[0152] u10 u component of wind speed at 10 m

[0153] v10 v component of wind speed at 10 m

[0154] wind_direction_100m 100m synthetic wind direction

[0155] wind_direction_10m 10m synthetic wind direction

[0156] wind_speed_100m 100m synthetic wind speed

[0157] wind_speed_10m 10m synthetic wind speed

[0158] Step 2: Model construction. Four machine learning models were selected to be integrated into the BMA model, namely Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and XGBoost models. During the hyperparameter adjustment process, 5-fold cross-validation was used to optimize model performance, and a grid search algorithm was used to explore different hyperparameter combinations for each model. Specific options are shown in Table 2. Finally, the highest R value in the validation set was selected. 2 Scores were obtained to screen out the optimal hyperparameter combination for each model. After obtaining the prediction results of the four machine learning models, the model weights were calculated based on the Bayesian Model Averaging (BMA) method using the differential evolution adaptive Metropolis Markov Chain Monte Carlo (MCMC) MATLAB toolbox developed by Jasper A. Vrugt, as shown in Equation (5). The weight combinations obtained by running the toolbox 20 times and the R values ​​of the average weight combinations were taken. 2 The highest weight combination W i =(w i,1 ,w i,2 ,w i,3 ,w i,4 ) is used as the optimal weight combination of the four machine learning models, and the BMA model prediction value is calculated based on this, as shown in formula (6).

[0159]

[0160]

[0161] Where: M j is the predicted value of the jth machine learning model; 4 is the number of machine learning models in this study; D = [y1, y2, ..., y N ] represents the observed data; P(D|M j ) is the predicted value M of the given model j The conditional probability of rainfall data, P(M j ) is the prior probability of the jth model; M i is the model prediction value of the i-th machine learning model; P(M i ) is the prior probability of the i-th machine learning model; P(M i |D) is the model prediction value M i The posterior probability of P(M j |D) is the model prediction value M j The posterior probability, w j is the model weight, satisfying pj (y|M j ,D) is the given model prediction M j and the posterior distribution of y given the observed data D. opt,j is the predicted value of the j-th machine learning model, and p(y|D) is the predicted value of the BMA model.

[0162] Table 2. Hyperparameter options for RF, SVM, KNN, and XGBoost models

[0163]

[0164] Step 3: Result evaluation. The evaluation is based on three evaluation indicators: coefficient of determination (R 2 ), mean absolute error (MAE), and root mean square error (RMSE) are used to evaluate the performance of the BMA model in this example, see Equations (7), (8), and (9).

[0165]

[0166]

[0167]

[0168] Where: and are the i-th rainfall observation and forecast values ​​respectively; is the mean of the observed values.

[0169] As an optional implementation method, the present invention uses Guangdong Province as the research area:

[0170] refer to Figure 2 In this case, a buffer zone with a radius of 400 kilometers was constructed based on the typhoon center through the grid data extraction method. The grids where the buffer zone overlaps with the study area were identified as the typhoon impact area and used as the target area for typhoon rainfall prediction. This study ultimately screened out 28 typical typhoon events, as follows: Figure 3 As shown, 28 typhoon paths and 462 typhoon centers were extracted for the period 2020-2023, of which the typhoon centers were obtained every 3 hours.

[0171] Compared with four individual machine learning models—RF, SVM, KNN, and XGBoost—the BMA model, using the method proposed in this example, is the most robust method for typhoon rainfall forecasting. By optimizing and weighting the prediction results of multiple models, BMA effectively leverages the strengths of each model while minimizing the risk of overfitting. While individual models often ignore structural uncertainty, BMA explicitly quantifies this uncertainty, resulting in more reliable and trustworthy forecasts. An evaluation of the prediction results of the BMA model and the four machine learning models is shown in Tables 3, 4, and 5.

[0172] Table 3. R2 of the BMA model and four machine learning models

[0173] Forecast period (h) 0 1 2 3 4 5 6 BMA 0.90 0.84 0.82 0.82 0.82 0.81 0.81 RF 0.86 0.75 0.75 0.76 0.77 0.76 0.77 Support Vector Machine 0.87 0.82 0.79 0.79 0.78 0.77 0.77 KNN 0.86 0.78 0.76 0.75 0.75 0.74 0.74 XGBoost 0.88 0.81 0.81 0.80 0.81 0.80 0.81

[0174] Table 4. MAE of the BMA model and four machine learning models

[0175] Forecast period (h) 0 1 2 3 4 5 6 BMA 0.30 0.38 0.40 0.40 0.43 0.43 0.42 RF 0.36 0.51 0.52 0.52 0.50 0.50 0.49 Support Vector Machine 0.31 0.39 0.41 0.42 0.42 0.44 0.44 KNN 0.36 0.43 0.45 0.46 0.47 0.49 0.48 XGBoost 0.34 0.43 0.45 0.45 0.46 0.46 0.46

[0176] Table 5. RMSE of the BMA model and four machine learning models

[0177] Forecast period (h) 0 1 2 3 4 5 6 BMA 0.61 0.76 0.80 0.81 0.81 0.83 0.83 RF 0.70 0.94 0.96 0.94 0.92 0.94 0.92 Support Vector Machine 0.68 0.81 0.86 0.88 0.89 0.93 0.91 KNN 0.71 0.88 0.93 0.95 0.95 0.99 0.97 XGBoost 0.65 0.81 0.84 0.84 0.83 0.85 0.84

[0178] The key points of the present invention are:

[0179] (1) Selection of model characteristic variables: This paper constructs a buffer zone with a radius of 400 kilometers based on the typhoon center as the typhoon's impact range. Three types of characteristics within the typhoon's impact range are comprehensively considered: typhoon characteristics, grid spatial properties, and meteorological characteristics. The dynamic weight of each grid in the study area is calculated based on its distance from the typhoon center, and the dynamic weighted typhoon characteristics are further calculated. The lag of the rainfall variable is calculated and selected based on the autocorrelation coefficient. Before running the model, all variables are rescaled using the Z-score standardization method.

[0180] (2) BMA model construction method: This paper selects four machine learning models (RF, SVM, KNN, and XGBoost) and integrates them into the BMA model. Five-fold cross-validation and GridSearch are used to adjust hyperparameters to optimize the performance of individual machine learning models. The BMA model prediction values ​​are weighted and calculated based on the Bayesian model averaging method to obtain a more robust and accurate typhoon rainfall forecast with a forecast period of 0-6 hours.

[0181] Advantages of the present invention:

[0182] (1) The key contribution of dynamically weighted typhoon characteristics: Typhoon characteristics play a vital role in typhoon rainfall forecasting. Among them, the dynamically weighted typhoon wind speed contributes the most to rainfall forecasting, reflecting the close correlation between wind intensity and rainfall in each grid within the typhoon-affected area. The dynamic weighting method can effectively capture and describe the law of wind speed changes in the typhoon-affected area, thereby significantly improving the accuracy and explanatory power of typhoon rainfall forecasting. Compared with traditional methods, the present invention further optimizes the prediction accuracy and improves the consistency and reliability of the results by considering the spatial variation of typhoon intensity.

[0183] (2) Advantages of the BMA model in typhoon rainfall forecasting: This paper introduces the Bayesian model averaging method (BMA), which quantifies the prediction uncertainty of the models by assigning weights to multiple machine learning models, thereby effectively integrating the advantages of different models. Compared with the traditional single optimal model, BMA not only significantly reduces the overall prediction error, but also reduces the conventional error and extreme prediction error, and improves the robustness of the prediction results. The application of BMA makes typhoon rainfall forecasting more stable, can effectively deal with model uncertainty and complex meteorological conditions, and improves the reliability of the forecast.

[0184] On the other hand, an embodiment of the present invention also provides a typhoon rainfall prediction system based on machine learning and BMA model, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, it implements the typhoon rainfall prediction method based on machine learning and BMA model as described above.

[0185] The processor and the memory can be connected via a bus or other means. The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0186] On the other hand, an embodiment of the present invention further provides a computer-readable storage medium, which stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the typhoon rainfall prediction method based on machine learning and BMA model as described above.

[0187] Those skilled in the art will appreciate that all or some of the steps and systems in the method disclosed above can be implemented as software, firmware, hardware, and appropriate combinations thereof. Some physical components or all physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, and the computer-readable medium can include computer storage media (or non-transitory media) and communication media (or temporary media). As known to those skilled in the art, the term computer storage media is included in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data) and is volatile and non-volatile, removable, and non-removable. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory, or other memory technology, CD-ROM, digital versatile disks (DVD), or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage, or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, as is well known to those skilled in the art, communication media typically embodies computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0188] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and essence of the present invention should be within the scope of the present invention.

Claims

1. A typhoon rainfall prediction method based on machine learning and BMA model, characterized in that: The typhoon rainfall prediction method based on machine learning and the BMA model includes the following steps: Acquiring characteristic information; the characteristic information includes typhoon characteristics, grid space attributes and meteorological characteristics; According to the feature information, a machine learning model is obtained; the machine learning model includes a RF model, a SVM model, a KNN model and an XGBoost model; According to the machine learning model, a BMA model is obtained; Typhoon rainfall forecast data is obtained according to the BMA model.

2. The typhoon rainfall prediction method based on machine learning and BMA model according to claim 1 is characterized in that: The step of obtaining characteristic information includes the following steps: Obtain typhoon center data; constructing a buffer zone based on the typhoon center data by using a grid data extraction method; Obtaining characteristic information according to the buffer zone; the characteristic information includes typhoon characteristics, grid space attributes and meteorological characteristics; The typhoon characteristics include the longitude of the typhoon center, the latitude of the typhoon center, the maximum sustained wind speed, the weighted typhoon wind speed, the typhoon center air pressure, the weighted typhoon center air pressure, the moving speed of the typhoon center, the weighted typhoon moving speed, and the moving direction of the typhoon center; the weighted typhoon wind speed, the weighted typhoon center air pressure, and the weighted typhoon moving speed are the dynamic weighted typhoon characteristics in the typhoon characteristics; the maximum sustained wind speed, the typhoon center air pressure, and the typhoon center moving speed are the original typhoon characteristics of the typhoon center in the typhoon characteristics; The grid spatial attributes include the grid latitude, grid longitude, elevation information, the distance of the grid from the typhoon center, and the orientation of the grid relative to the typhoon center of each grid; The meteorological characteristics include rainfall, temperature, relative humidity, wind speed and wind direction of each grid; The lag time of the rainfall is selected by the autocorrelation coefficient; The formula used to calculate the dynamic weighted typhoon characteristics in each grid includes: oh i =1-d i / r; Weighted_f i =f·ω i ; Where i is a natural number: d i is the distance between the ith grid and the typhoon center; r is the radius of the buffer zone; ω i is the dynamic weight of the i-th grid; f represents the original typhoon characteristics; Weighted_f i is the dynamic weighted typhoon characteristic of the i-th grid.

3. The typhoon rainfall prediction method based on machine learning and BMA model according to claim 2 is characterized in that: The lag time of the rainfall is selected by the autocorrelation coefficient, and the formula used includes: Where N is the total number of observations in the time series; t is the first moment, tk is the second moment; R t,i is the rainfall of the grid at the first moment; R t-k,i is the rainfall of the grid at the second moment; is the average rainfall of the grid at the first moment; is the average rainfall of the grid at the second moment; k is an integer, 1≤k≤24; ρ(k) is the lag time of rainfall.

4. The typhoon rainfall prediction method based on machine learning and BMA model according to claim 1, characterized in that: The step of obtaining a machine learning model based on the feature information includes the following steps: The feature information is processed using a Z-score normalization method, and each feature information is rescaled so that all the feature information has a zero mean and a unit standard deviation, thereby obtaining model training data; Training an initial training model according to the model training data; the initial training model includes an initial RF model, an initial SVM model, an initial KNN model and an initial XGBoost model; Evaluate the performance of each of the initial training models through five-fold cross validation, and use a grid search algorithm to perform hyperparameter tuning to obtain the optimal hyperparameter combination for each of the initial training models; According to the optimal hyperparameter combination, a machine learning model is obtained; the machine learning model includes an RF model, an SVM model, a KNN model and an XGBoost model.

5. The typhoon rainfall prediction method based on machine learning and BMA model according to claim 1, characterized in that: The step of obtaining a BMA model according to the machine learning model includes the following steps: Obtaining prediction results of each of the machine learning models; Obtaining a model weight according to the prediction result; According to the model weights, a BMA model is obtained.

6. The typhoon rainfall prediction method based on machine learning and BMA model according to claim 5, characterized in that: The model weight is obtained according to the prediction result, and the formula used includes: Among them, M j is the model prediction value of the j-th machine learning model; D = [y1, y2, ..., y N ] represents the observed data; P(D|M j ) is the predicted value M of the given model j The conditional probability of rainfall data; P(M j ) is the prior probability of the j-th machine learning model; M i is the model prediction value of the i-th machine learning model; P(M i ) is the prior probability of the i-th machine learning model; P(M i |D) is the model prediction value M i The posterior probability of P(M j |D) is the model prediction value M j The posterior probability, w j is the model weight, satisfying 7. The typhoon rainfall prediction method based on machine learning and BMA model according to claim 1, characterized in that: The method further comprises the following steps: Get rainfall observations; Get the predicted value of the BMA model; Evaluation data is obtained based on the rainfall observation value and the predicted value; the evaluation data includes a determination coefficient, a mean absolute error, and a root mean square error.

8. The typhoon rainfall prediction method based on machine learning and BMA model according to claim 7, characterized in that: The evaluation data is obtained based on the rainfall observation value and the predicted value, and the formula used includes: in, is the i-th rainfall observation value; is the i-th predicted value; is the average value of the rainfall observations; R 2 is the coefficient of determination; MAE is the mean absolute error; RMSE is the root mean square error.

9. A typhoon rainfall prediction system based on machine learning and BMA model, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for predicting typhoon rainfall based on machine learning and the BMA model as claimed in any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the typhoon rainfall prediction method based on machine learning and BMA model as described in any one of claims 1 to 8.

Images