A method for classifying sunny and rainy periods based on multi-time-scale combination of microwave attenuation data

Abstract

This invention provides a method for classifying rainy seasons using microwave signals across multiple time scales, relating to the field of precipitation monitoring technology. The method includes acquiring microwave attenuation signal intensity data and rainfall data from ground stations within a target watershed; data preprocessing; selecting and statistically analyzing the feature values ​​of each microwave attenuation signal at different time scales to form a feature matrix, which is then used as the dataset for a machine learning model; selecting the first 75% of the dataset as the training set to train machine learning models at different time scales; determining the optimal parameters through Bayesian optimization; selecting the last 25% of the dataset as the validation set to validate the machine learning models at different time scales; using measured data from the target watershed to drive the validated models, obtaining discrimination results at different time scales, and using a weighted factor scoring method to determine the rainy season. This method integrates feature information from multiple time scales, overcoming the limitations of single-time-scale discrimination and improving the reliability of rainy season discrimination.

Description

Technical Field

[0001] This invention relates to the field of precipitation monitoring technology, and in particular to a method for classifying sunny and rainy periods by combining microwave signals at multiple time scales. Background Technology

[0002] To effectively improve the early warning level of precipitation monitoring, it is necessary to enhance the accuracy and real-time performance of precipitation monitoring. Currently, commonly used precipitation monitoring methods include the following three types: rain gauges, weather radar, and rain-measuring satellites. Although these methods are widely used in the water resources and meteorological industry, they have significant limitations, such as easily missing the center of heavy rainfall, limited effectiveness in converting "clouds in the air" into actual "ground rain," low inversion accuracy, and poor real-time performance.

[0003] The method of classifying sunny and rainy periods using the rain attenuation effect of microwave links is an important part of microwave link precipitation monitoring and the basis for inverting track-based rainfall intensity. However, most existing methods for classifying sunny and rainy periods based on the rain attenuation effect of microwave links focus on classification problems at a single time scale. They are often limited by the selection of representative time scales, fail to consider the different characteristic information of data at different time scales, and do not fully utilize the advantages of multi-scale features of the data, thus having shortcomings in combining multi-scale features.

[0004] Therefore, it is necessary to provide a method for classifying sunny and rainy seasons by combining multiple time scales of microwave signals to solve the above-mentioned technical problems. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method for classifying sunny and rainy periods using a combination of microwave signals across multiple time scales. This method addresses the limitations of existing methods for classifying sunny and rainy periods based on the rain attenuation effect of microwave links, which mostly focus on classification at a single time scale. These methods are constrained by the selection of representative time scales and cannot further explore the advantages of microwave attenuation data at multiple time resolutions, thus failing to achieve the problem of classifying sunny and rainy periods based on a combination of multi-scale features.

[0006] This invention provides a method for classifying sunny and rainy periods using a multi-timescale combination of microwave signals, comprising:

[0007] Acquire microwave attenuation signal data and rainfall data from ground stations in the target watershed;

[0008] The microwave attenuation signal data is preprocessed;

[0009] Feature values ​​of preprocessed microwave attenuation signal data at different time scales are selected and statistically analyzed to determine the corresponding feature matrix. Combined with the rainfall data, the input dataset for the machine learning model is constructed.

[0010] The input dataset is divided into a training set and a validation set, and machine learning models at different time scales are trained based on the training set.

[0011] The optimal parameters of machine learning models at different time scales are determined using the Bayesian optimization algorithm.

[0012] The effectiveness of machine learning models at different time scales was verified based on the validation set.

[0013] Based on the microwave attenuation signal data, the rainfall period discrimination results of machine learning models at different time scales are determined, and the final rainfall period discrimination result is determined by the weighted factor scoring method based on the rainfall period discrimination results.

[0014] Preferably, microwave attenuation signal data of the target watershed at 15-minute and 1-hour resolutions are acquired using a microwave signal transceiver device.

[0015] Preferably, the rainfall data from the ground stations includes measured rainfall data from rain gauge stations, meteorological stations, and hydrological stations.

[0016] Preferably, the preprocessing of the microwave attenuation signal data includes removing the microwave attenuation signal data corresponding to abnormal microwave links and performing clutter filtering on the microwave attenuation signal data corresponding to valid microwave links.

[0017] Preferably, the step of selecting and statistically analyzing the feature values ​​of preprocessed microwave attenuation signal data at different time scales, determining the corresponding feature matrix, and constructing the input dataset for the machine learning model in conjunction with the rainfall data specifically includes:

[0018] Feature values ​​of preprocessed microwave attenuation signal data at different time scales were selected and statistically analyzed to construct the corresponding feature matrix. ,in, , , , , , These represent the microwave link length, microwave attenuation signal frequency, microwave attenuation signal mean, microwave attenuation signal median, microwave attenuation signal maximum value, and microwave attenuation signal minimum value, respectively; k represents the sequence number of the microwave attenuation signal data. ;

[0019] Determine the label value based on the rainfall data. Based on the feature matrix and the label values, the input dataset for the machine learning model is constructed. And determine the datasets Q1 and Q2 corresponding to 15-minute and 1-hour resolutions, where the label values A value of 0 indicates a non-rainy period; the label value is... A value of 1 indicates a rainy season.

[0020] Preferably, based on the training set corresponding to the dataset Q1, the parameters of the support vector machine model are set, and the Gaussian radial basis function kernel function is selected as the kernel function of the support vector machine model. The calculation formula of the Gaussian radial basis function kernel function is as follows:

[0021] In the formula, Represents the Gaussian radial basis kernel function; Representation of the characteristic matrix With tag value Euclidean distance; This represents the width of the Gaussian radial basis kernel function. ; This represents an exponential function with the natural constant e as its base.

[0022] Based on the training set corresponding to the dataset Q2, the parameters of the category boosting model are set, and a data pool is created.

[0023] Preferably, the parameters of the category boosting model include the loss function, decision tree depth, learning rate, proportion of subsamples trained in each iteration, feature sampling proportion at each level, maximum number of iterations, and number of rounds before early termination.

[0024] Preferably, determining the optimal parameters of the machine learning model at different time scales using the Bayesian optimization algorithm specifically includes:

[0025] Obtain the measured rainfall period results and the rainfall period prediction results of the machine learning model. Set the confusion matrix corresponding to the machine learning model, including true positives (TP), true negatives (TN), false positives (FP), and false negatives (FN). True positives (TP) indicate that both the measured rainfall period results and the rainfall period prediction results of the machine learning model are wet periods; true negatives (TN) indicate that both the measured rainfall period results and the rainfall period prediction results of the machine learning model are dry periods; false positives (FP) indicate that the measured rainfall period results are dry periods, and the rainfall period prediction results of the machine learning model are wet periods; false negatives (FN) indicate that the measured rainfall period results are wet periods, and the rainfall period prediction results of the machine learning model are dry periods.

[0026] The precision of the machine learning model is calculated based on the confusion matrix, and the precision is used as the objective function. The formula for calculating the precision is as follows:

[0027] In the formula, ACC represents precision; TP represents true positives; TN represents true negatives; FP represents false positives; and FN represents false negatives.

[0028] The target space of the Bayesian optimization algorithm is set, and a uniform distribution is selected as the prior distribution of the Bayesian optimization algorithm to maximize the accuracy and determine the optimal parameters of the machine learning model.

[0029] Preferably, the step of determining the rainfall period model discrimination results of machine learning models at different time scales based on the microwave attenuation signal data, and determining the final rainfall period discrimination result based on the rainfall period model discrimination results using a weighted factor scoring method, specifically includes:

[0030] The microwave attenuation signal data is input into the validated support vector machine model, and the result of the rainfall period model is output at a 15-minute resolution. The microwave attenuation signal data is also input into the validated class boosting model, and the result of the rainfall period model is output at a 1-hour resolution.

[0031] Setting each hour as a statistical period, the support vector machine model outputs the rainfall period discrimination result every 15 minutes. , , , The classification improvement model outputs the model discrimination result for each hour of rainfall as follows: The total score f is calculated using a weighted factor scoring method, and the corresponding calculation formula is as follows:

[0032] In the formula, f represents the total score; Indicates the weight score; This represents the model's discrimination result for the rainfall period, where the rainfall period model discrimination result represents the non-rainfall period. A value of 0 indicates that the rainfall period is determined by the model. The value is 1;

[0033] like If the result is positive, the final rainfall period is determined to be a rainfall period; otherwise, the final rainfall period is determined to be a non-rainy period.

[0034] Preferably, the weighting , , , , The values ​​are 0.1, 0.15, 0.2, 0.25, and 0.3.

[0035] Compared with related technologies, the method for classifying sunny and rainy seasons by combining multiple time scales of microwave signals provided by this invention has the following beneficial effects:

[0036] This invention acquires microwave attenuation signal data and rainfall data from ground stations in a target watershed; preprocesses the microwave attenuation signal data; selects and statistically analyzes the feature values ​​of the preprocessed microwave attenuation signal data at different time scales to determine the corresponding feature matrix, and constructs the input dataset for a machine learning model by combining it with rainfall data; divides the input dataset into a training set and a validation set, and trains machine learning models at different time scales based on the training set; determines the optimal parameters of the machine learning models at different time scales using a Bayesian optimization algorithm; validates the performance of the machine learning models at different time scales based on the validation set; determines the rainfall period discrimination results of the machine learning models at different time scales based on the microwave attenuation signal data, and uses a weighted factor scoring method to determine the final rainfall period discrimination result based on the rainfall period discrimination results. This allows for further exploration of the advantages of microwave attenuation data at multiple time resolutions, integrates feature information from multiple time scales, overcomes the limitations of single-time-scale discrimination, and improves the reliability of rainfall period discrimination. This enables the classification of sunny and rainy periods based on multi-scale feature combinations, solving the problem that most existing methods focus on classification at a single time scale and are limited by the selection of representative time scales. Furthermore, this invention's method has wide applications in the field of microwave signal inversion for rainfall retrieval. Attached Figure Description

[0037] Figure 1 This is a flowchart of a method for classifying sunny and rainy periods using a multi-timescale combination of microwave signals according to the present invention;

[0038] Figure 2 This is a distributed packet graph representing the verification results of the present invention.

[0039] Figure 3 This is a diagram showing the results of the rainy / sunny season determination according to the present invention. Detailed Implementation

[0040] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0041] like Figure 1 As shown, a method for classifying sunny and rainy seasons using microwave signals with multi-timescale combination discrimination includes:

[0042] S1, acquire microwave attenuation signal data of the target watershed and rainfall data from ground stations;

[0043] S2, preprocess the microwave attenuation signal data;

[0044] S3. Select and statistically analyze the feature values ​​of the preprocessed microwave attenuation signal data at different time scales, determine the corresponding feature matrix, and combine it with the rainfall data to construct the input dataset for the machine learning model.

[0045] S4, the input dataset is divided into a training set and a validation set, and machine learning models at different time scales are trained based on the training set;

[0046] S5, using the Bayesian optimization algorithm to determine the optimal parameters of machine learning models at different time scales;

[0047] S6. Validate the performance of machine learning models at different time scales based on the validation set.

[0048] S7. Based on the microwave attenuation signal data, determine the rainfall period model discrimination results of machine learning models at different time scales, and use the weight factor scoring method to determine the final rainfall period discrimination result based on the rainfall period model discrimination results.

[0049] It should be noted that the research area of ​​this invention is located in a mountainous area near the dam in Yichang City, belonging to the mountainous landform of the Three Gorges of the Yangtze River. The basin exhibits significant elevation differences, with undulating mountains and numerous valleys in the upper and middle reaches, posing a risk of rockfalls; the lower reaches widen to tens of meters, consisting mainly of open piedmont plains with many villages and farmland. This basin has a subtropical continental monsoon climate. Since the establishment of the representative hydrological station in this basin, two major floods have occurred, with the peak flow at the controlled section reaching as high as 732 m³ / h. 3 / s, and the basin faces the risk of sudden localized torrential rains and floods.

[0050] In the specific implementation process, microwave attenuation signal data of the target watershed at 15-minute and 1-hour resolutions are acquired based on microwave signal transceivers.

[0051] In practical applications, based on the 50 microwave signal transceivers deployed in the basin, microwave attenuation signal data of each microwave link at 15-minute and 1-hour resolutions can be obtained.

[0052] The rainfall data from the ground stations includes measured rainfall data from rain gauge stations, meteorological stations, and hydrological stations.

[0053] In practical applications, the rainfall data from ground stations specifically includes measured rainfall data from 18 newly built meteorological stations and 10 existing hydrological stations within the basin.

[0054] The preprocessing of the microwave attenuation signal data includes removing the microwave attenuation signal data corresponding to abnormal microwave links and performing clutter filtering on the microwave attenuation signal data corresponding to valid microwave links.

[0055] The preprocessing of microwave attenuation signal data specifically includes: removing abnormal microwave links with obvious abnormal fluctuations, excessive extreme values, and large areas of missing data, and filtering clutter from the effective microwave links.

[0056] The process of selecting and statistically analyzing the feature values ​​of preprocessed microwave attenuation signal data at different time scales, determining the corresponding feature matrix, and combining this with the rainfall data to construct the input dataset for the machine learning model specifically includes:

[0057] Feature values ​​of preprocessed microwave attenuation signal data at different time scales were selected and statistically analyzed to construct the corresponding feature matrix. ,in, , , , , , These represent the microwave link length, microwave attenuation signal frequency, microwave attenuation signal mean, microwave attenuation signal median, microwave attenuation signal maximum value, and microwave attenuation signal minimum value, respectively; k represents the sequence number of the microwave attenuation signal data. n=100;

[0058] Determine the label value based on the rainfall data. Based on the feature matrix and the label values, the input dataset for the machine learning model is constructed. And determine the datasets Q1 and Q2 corresponding to 15-minute and 1-hour resolutions, where the label values A value of 0 indicates a non-rainy period; the label value is... A value of 1 indicates a rainy season.

[0059] Based on the training set corresponding to dataset Q1, the parameters of the support vector machine model are set, and the Gaussian radial basis function kernel function is selected as the kernel function of the support vector machine model. The calculation formula of the Gaussian radial basis function kernel function is as follows:

[0060] In the formula, Represents the Gaussian radial basis kernel function; Representation of the characteristic matrix With tag value Euclidean distance; This represents the width of the Gaussian radial basis kernel function. ; This represents an exponential function with the natural constant e as its base.

[0061] Based on the training set corresponding to the dataset Q2, the parameters of the category boosting model are set, and a data pool is created.

[0062] It should be noted that when training the Support Vector Machine (SVM) model, the first 75% of the data in dataset Q1 can be used as the training set, the parameters of the SVM model can be set, and the Gaussian radial basis kernel function can be selected as the kernel function of the model to determine the rainfall period.

[0063] When training the CatBoost model, the first 75% of the data in dataset Q2 can be used as the training set, i.e., the data from 0:00 on May 1 to 24:00 on July 31. The parameters of the CatBoost model are set, the Pool class is created, and then the rainfall period is determined.

[0064] The parameters of the category boosting model include the loss function, decision tree depth, learning rate, subsample ratio for training in each iteration, feature sampling ratio for each level, maximum number of iterations, and number of rounds to stop early.

[0065] The parameters of the category boosting model include: loss function (loss_function), value Logloss; decision tree depth (depth), value 8; learning rate (learning_rate), value 0.04; subsample ratio for each training iteration (subsample), value 0.8; colsample_bylevel ratio for each level of features (colsample_bylevel), value 0.8; maximum number of iterations (iterations), value 1500; and early stopping rounds (early_stopping_rounds), value 20.

[0066] The determination of optimal parameters for machine learning models at different time scales using the Bayesian optimization algorithm specifically includes:

[0067] As shown in Table 1, the measured rainfall period results and the rainfall period prediction results of the machine learning model are obtained. A confusion matrix corresponding to the machine learning model is set, including true positives (TP), true negatives (TN), false positives (FP), and false negatives (FN). Here, true positives (TP) indicate that both the measured rainfall period results and the rainfall period prediction results of the machine learning model are wet periods; true negatives (TN) indicate that both the measured rainfall period results and the rainfall period prediction results of the machine learning model are dry periods; false positives (FP) indicate that the measured rainfall period results are dry periods, and the rainfall period prediction results of the machine learning model are wet periods; false negatives (FN) indicate that the measured rainfall period results are wet periods, and the rainfall period prediction results of the machine learning model are dry periods.

[0068] Table 1. Confusion Matrix Element Table

[0069]

[0070] The precision of the machine learning model is calculated based on the confusion matrix, and the precision is used as the objective function. The formula for calculating the precision is as follows:

[0071] In the formula, ACC represents precision; TP represents true positives; TN represents true negatives; FP represents false positives; and FN represents false negatives.

[0072] The target space of the Bayesian optimization algorithm is set, and a uniform distribution is selected as the prior distribution of the Bayesian optimization algorithm to maximize the accuracy and determine the optimal parameters of the machine learning model.

[0073] In practical applications, the last 25% of the input dataset is selected as the validation set, specifically the data from 00:00 on August 1st to 24:00 on August 31st. The effectiveness of the machine learning model at different time scales is then validated based on this validation set. The distributed matrix graph of the validation results is shown below. Figure 2 As shown.

[0074] Based on the microwave attenuation signal data, the rainfall period discrimination results of machine learning models at different time scales are determined, and the final rainfall period discrimination result is determined using a weighted factor scoring method based on the rainfall period discrimination results. Specifically, this includes:

[0075] The microwave attenuation signal data is input into the validated support vector machine model, and the result of the rainfall period model is output at a 15-minute resolution. The microwave attenuation signal data is also input into the validated class boosting model, and the result of the rainfall period model is output at a 1-hour resolution.

[0076] Setting each hour as a statistical period, the support vector machine model outputs the rainfall period discrimination result every 15 minutes. , , , The classification improvement model outputs the model discrimination result for each hour of rainfall as follows: The total score f is calculated using a weighted factor scoring method, and the corresponding calculation formula is as follows:

[0077] In the formula, f represents the total score; Indicates the weight score; This represents the model's discrimination result for the rainfall period, where the rainfall period model discrimination result represents the non-rainfall period. A value of 0 indicates that the rainfall period is determined by the model. The value is 1;

[0078] like If the result is positive, the final rainfall period is determined to be a rainfall period; otherwise, the final rainfall period is determined to be a non-rainy period. The corresponding determination results are as follows: Figure 3 As shown.

[0079] The weighting , , , , The values ​​are 0.1, 0.15, 0.2, 0.25, and 0.3.

[0080] Through the above embodiments, this invention provides a method for classifying sunny and rainy periods using a multi-timescale combination of microwave signals. This method acquires microwave attenuation signal data from the target watershed and rainfall data from ground stations; preprocesses the microwave attenuation signal data; selects and statistically analyzes the feature values ​​of the preprocessed microwave attenuation signal data at different time scales to determine the corresponding feature matrix; and combines this with rainfall data to construct the input dataset for a machine learning model. The input dataset is divided into a training set and a validation set. Machine learning models at different time scales are trained based on the training set. The optimal parameters of the machine learning models at different time scales are determined using a Bayesian optimization algorithm. Finally, the machine learning models at different time scales are validated based on the validation set. The effectiveness of the model is demonstrated by determining the rainfall period discrimination results of machine learning models at different time scales based on microwave attenuation signal data. A weighted factor scoring method is then used to determine the final rainfall period discrimination result based on these results. This allows for further exploration of the advantages of microwave attenuation data at multiple time resolutions, integrating feature information from multiple time scales to overcome the limitations of single-time-scale discrimination and improve the reliability of rainfall period discrimination. This enables the classification of sunny and rainy periods based on multi-scale feature combinations, solving the problem that most existing methods focus on classification at a single time scale and are limited by the selection of representative time scales. Furthermore, this invention has wide applications in the field of rainfall retrieval from microwave signals.

[0081] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0082] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0083] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

Claims

1. A method for classifying sunny and rainy periods using microwave signals with multi-timescale combination discrimination, characterized in that, include: Acquire microwave attenuation signal data and rainfall data from ground stations in the target watershed; The microwave attenuation signal data is preprocessed; Feature values ​​of preprocessed microwave attenuation signal data at different time scales are selected and statistically analyzed to determine the corresponding feature matrix. Combined with the rainfall data, the input dataset for the machine learning model is constructed. The input dataset is divided into a training set and a validation set, and machine learning models at different time scales are trained based on the training set. The optimal parameters of machine learning models at different time scales are determined using the Bayesian optimization algorithm. The effectiveness of machine learning models at different time scales was verified based on the validation set. Based on the microwave attenuation signal data, the rainfall period model discrimination results of machine learning models at different time scales are determined, and the final rainfall period discrimination result is determined based on the rainfall period model discrimination results using the weighted factor scoring method. The process of selecting and statistically analyzing the feature values ​​of preprocessed microwave attenuation signal data at different time scales, determining the corresponding feature matrix, and combining this with the rainfall data to construct the input dataset for the machine learning model specifically includes: Feature values ​​of preprocessed microwave attenuation signal data at different time scales were selected and statistically analyzed to construct the corresponding feature matrix. in, These represent the microwave link length, microwave attenuation signal frequency, microwave attenuation signal mean, microwave attenuation signal median, microwave attenuation signal maximum value, and microwave attenuation signal minimum value, respectively; k represents the sequence number of the microwave attenuation signal data, k = 1, 2, ..., n; Determine the label value y based on the rainfall data. k Based on the feature matrix and the label values, construct the input dataset Q = {(x k ,y k )}, and determine the datasets Q1 and Q2 corresponding to 15-minute and 1-hour resolutions, where the label value y k A value of 0 indicates a non-rainy period, and the label value is y. k A value of 1 indicates a rainy season; Based on the training set corresponding to dataset Q1, the parameters of the support vector machine model are set, and the Gaussian radial basis function kernel function is selected as the kernel function of the support vector machine model. The calculation formula of the Gaussian radial basis function kernel function is as follows: In the formula, K(x) k ,y k ) represents the Gaussian radial basis function; ||x k -y k || represents the characteristic matrix x k With label value y k The Euclidean distance; σ represents the width of the Gaussian radial basis kernel function, σ>0; exp() represents the exponential function with the natural constant e as the base; Based on the training set corresponding to the dataset Q2, the parameters of the category boosting model are set, and a data pool is created.

2. The method for classifying sunny and rainy periods by combining microwave signals at multiple time scales according to claim 1, characterized in that, Based on the microwave signal transceiver device, microwave attenuation signal data of the target watershed at 15-minute and 1-hour resolutions are acquired for each microwave link.

3. The method for classifying sunny and rainy periods by combining microwave signals at multiple time scales according to claim 1, characterized in that, The rainfall data from the ground stations includes measured rainfall data from rain gauge stations, meteorological stations, and hydrological stations.

4. The method for classifying sunny and rainy periods by combining microwave signals across multiple time scales according to claim 1, characterized in that, The preprocessing of the microwave attenuation signal data includes removing the microwave attenuation signal data corresponding to abnormal microwave links and performing clutter filtering on the microwave attenuation signal data corresponding to valid microwave links.

5. The method for classifying sunny and rainy periods by combining microwave signals at multiple time scales according to claim 1, characterized in that, The parameters of the category boosting model include the loss function, decision tree depth, learning rate, subsample ratio for training in each iteration, feature sampling ratio for each level, maximum number of iterations, and number of rounds to stop early.

6. The method for classifying sunny and rainy periods using a multi-timescale combination of microwave signals according to claim 1, characterized in that, The determination of optimal parameters for machine learning models at different time scales using the Bayesian optimization algorithm specifically includes: Obtain the measured rainfall period results and the rainfall period prediction results of the machine learning model. Set the confusion matrix corresponding to the machine learning model, including true positives (TP), true negatives (TN), false positives (FP), and false negatives (FN). True positives (TP) indicate that both the measured rainfall period results and the rainfall period prediction results of the machine learning model are wet periods; true negatives (TN) indicate that both the measured rainfall period results and the rainfall period prediction results of the machine learning model are dry periods; false positives (FP) indicate that the measured rainfall period results are dry periods, and the rainfall period prediction results of the machine learning model are wet periods; false negatives (FN) indicate that the measured rainfall period results are wet periods, and the rainfall period prediction results of the machine learning model are dry periods. The precision of the machine learning model is calculated based on the confusion matrix, and the precision is used as the objective function. The formula for calculating the precision is as follows: In the formula, ACC represents precision; TP represents true positives; TN represents true negatives; FP represents false positives; and FN represents false negatives. The target space of the Bayesian optimization algorithm is set, and a uniform distribution is selected as the prior distribution of the Bayesian optimization algorithm to maximize the accuracy and determine the optimal parameters of the machine learning model.

7. The method for classifying sunny and rainy periods by combining microwave signals at multiple time scales according to claim 1, characterized in that, Based on the microwave attenuation signal data, the rainfall period discrimination results of machine learning models at different time scales are determined, and the final rainfall period discrimination result is determined using a weighted factor scoring method based on the rainfall period discrimination results. Specifically, this includes: The microwave attenuation signal data is input into the validated support vector machine model, and the result of the rainfall period model is output at a 15-minute resolution. The microwave attenuation signal data is also input into the validated class boosting model, and the result of the rainfall period model is output at a 1-hour resolution. The statistical period is set to 1 hour. The model discrimination results for the rainfall period output by the support vector machine model for every 15 minutes are r1, r2, r3, and r4. The model discrimination result for the rainfall period output by the category boosting model for every hour is r5. The total score f is calculated using the weighted factor scoring method. The corresponding calculation formula is as follows: In the formula, f represents the total score; λ i Indicates the weight score; r i This represents the model's discrimination result for the rainfall period, where the rainfall period model discrimination result represents the non-rainfall period, r. i When the value is 0, the model's determination of the rainfall period is the rainfall period, r i The value is 1; If f ≥ 0.6, then the final rainfall period determination result is determined to be a rainfall period; otherwise, the final rainfall period determination result is determined to be a non-rainfall period.

8. The method for classifying sunny and rainy periods using a multi-timescale combination of microwave signals according to claim 7, characterized in that, The values ​​of the weights λ1, λ2, λ3, λ4, and λ5 are 0.1, 0.15, 0.2, 0.25, and 0.3, respectively.

Images