A system and method for predicting high wind disaster weather index based on deep learning

Abstract

The present invention provides a system and method for predicting a gale disaster weather index based on deep learning, which relates to the technical field of weather monitoring, and specifically includes: collecting historical meteorological data of an area to be evaluated and the affected area and duration of gale disaster events occurring at a corresponding time; establishing an index prediction model based on a long short-term memory network, inputting the historical meteorological data into the model as a training set, and training the model using the corresponding affected area and duration of gale disasters as labels; collecting meteorological data of the area to be evaluated in real time, and inputting the data into a trained index prediction model to obtain a predicted affected area and a predicted duration; collecting terrain height and slope data to correct the predicted affected area and the predicted duration to obtain a comprehensive weather index, and comparing the index with a preset threshold, dividing the risk level according to the result, thereby improving the accuracy of gale disaster prediction and providing more scientific and effective decision support for disaster prevention and emergency management.

Description

Technical Field

[0001] The present invention relates to the field of weather monitoring technology, and in particular to a system and method for predicting a gale disaster weather index based on deep learning. Background Art

[0002] Strong wind disasters pose a serious threat to human society and the economy, so it is crucial to accurately predict the impact of strong wind disasters. Traditional weather forecasting methods mainly rely on numerical weather forecast models and empirical formulas, which cannot fully capture the impact of complex meteorological elements and terrain on strong wind disasters. In recent years, deep learning technology, especially long short-term memory networks, has performed well in processing time series data and can effectively capture the time dependence and nonlinear characteristics in meteorological data. Therefore, applying LSTM to the prediction of strong wind disaster weather index can improve the accuracy of the prediction.

[0003] With climate change, the frequency and intensity of extreme weather events have increased, which has put forward higher requirements for the accurate prediction of strong wind disasters. However, the prediction ability of traditional prediction methods is limited when faced with multivariable, multi-scale, and nonlinear meteorological data. Especially in areas with complex terrain, wind distribution is significantly affected by factors such as terrain height and slope. Traditional models often find it difficult to accurately reflect these influences, resulting in a large deviation between the prediction results and the actual disaster situation. The rasterization method is used to collect terrain height and slope data of the area to be evaluated. Through the correction of terrain data, the regulating effect of terrain on strong wind disasters is fully considered, making the prediction results more practical. In addition, with the development of sensor technology and data acquisition technology, more abundant and detailed meteorological data can be obtained, but how to effectively use these data and improve the performance of prediction models is an important challenge facing the current field of meteorological prediction.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0005] The purpose of the present invention is to provide a system and method for predicting a high wind disaster weather index based on deep learning to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A high wind disaster weather index prediction system based on deep learning, specifically including:

[0008] A data collection module is used to collect the affected area and duration of the strong wind disaster events that occurred in the history of the area to be evaluated, and collect the historical meteorological data of the strong wind disaster events in the area to be evaluated, which is traced back to one week ago. The meteorological data includes wind speed, humidity, temperature, air pressure, precipitation and cloud cover;

[0009] The model building module builds an index prediction model based on the long short-term memory network, inputs historical meteorological data as a training set into the index prediction model, uses the corresponding disaster area and duration of the gale disaster as labels, trains the model, and obtains a trained index prediction model;

[0010] The prediction module collects the meteorological data of the area to be evaluated in the past week and inputs it into the trained index prediction model to obtain the predicted disaster area and duration of the strong wind disaster weather. The terrain height and slope data of the area to be evaluated are collected using the rasterization method.

[0011] The comprehensive assessment module is used to use the collected terrain height and slope data of the area to be assessed to correct the predicted disaster-stricken area and predicted duration of strong wind disaster weather, obtain a comprehensive weather index, compare the comprehensive weather index with a preset assessment threshold, and divide the risk level according to the comparison results.

[0012] Furthermore, the specific logic for collecting the wind speed, humidity, temperature and air pressure in the area to be assessed one week before the occurrence of a strong wind disaster event in history is as follows:

[0013] The weather data recorded in the history of the Meteorological Bureau is used to obtain the wind speed, humidity, temperature and air pressure in the week before the strong wind disaster event occurred in the area to be assessed;

[0014] Select n measurement points around the area to be evaluated, covering the four directions of east, west, south and north, and select measurement points in equal quantities and distances in each direction;

[0015] The specific logic for collecting meteorological data for the area to be evaluated in the past week is as follows:

[0016] Install anemometers, hygrometers, thermometers and barometers at the measurement points;

[0017] Record wind speed data, humidity data, temperature data and air pressure data at different measurement points to obtain the average wind speed, average humidity, average temperature and average air pressure of the area to be evaluated;

[0018] The formula for obtaining the average wind speed is:

[0019]

[0020] in, is the average wind speed, v irepresents the wind speed measured at the i-th measurement point;

[0021] The formula used to obtain the average humidity is:

[0022]

[0023] in, is the average humidity, RH i represents the humidity measured at the i-th measurement point;

[0024] The formula used to obtain the average temperature is:

[0025]

[0026] in, is the average temperature, Temp i represents the temperature measured at the i-th measurement point;

[0027] The formula for obtaining the average air pressure is:

[0028]

[0029] in, is the average air pressure, PRES i represents the air pressure measured at the i-th measurement point.

[0030] Furthermore, the specific logic of collecting the terrain height H and slope data C of the area to be evaluated by using the rasterization method is:

[0031] The area to be evaluated is divided into regular small grids according to its size, and then the average height and slope are calculated in each grid unit to generate a continuous terrain surface model. Each unit represents an average value, and the heights and slopes of all grid units are weighted averaged again. The final values ​​are used as the terrain height H and slope C of the area to be evaluated.

[0032] Furthermore, the specific logic for collecting the precipitation M and cloud cover N in the week before the strong wind disaster event occurred in the history of the area to be evaluated is:

[0033] The historical weather data recorded by the Meteorological Bureau are used to obtain the daily average precipitation M and daily average cloud cover N in the week before the strong wind disaster event in the area to be assessed;

[0034] The collected cloud amount N is the proportion of the sky in the disaster area covered by clouds when a strong wind disaster occurs, expressed as a percentage, ranging from 0% to 100%;

[0035] The specific logic for collecting precipitation and cloud cover data for the area to be evaluated in the past week is as follows:

[0036] The precipitation and cloud cover in the area to be evaluated in the past week are collected through the historical weather data recorded by the Meteorological Bureau and the prediction results of the weather forecast.

[0037] Furthermore, the process of establishing an index prediction model based on the long short-term memory network specifically includes:

[0038] Historical meteorological data are organized into multiple groups of time series data according to their types and used as training sets. The affected area and duration of the strong wind disasters are used as labels. A deep learning model based on the long short-term memory network is constructed. The training set and labels are input into the deep learning model, and the model is trained to obtain a trained index prediction model. The historical meteorological data include wind speed, humidity, temperature, air pressure, precipitation and cloud cover in the area to be evaluated.

[0039] Furthermore, the specific logic for obtaining the predicted disaster-affected area S and predicted duration T of the strong wind disaster weather is as follows:

[0040] The meteorological data of the area to be evaluated in the past week are collected, including wind speed, humidity, temperature, air pressure, precipitation and cloud cover in the area to be evaluated, and input into the trained exponential prediction model to output the predicted disaster-stricken area S and predicted duration T of the strong wind disaster weather.

[0041] Furthermore, the terrain height and slope data of the area to be evaluated are used to correct the predicted disaster area and predicted duration of strong wind weather, and the formula for the comprehensive weather index is as follows:

[0042]

[0043] Among them, QS is the comprehensive weather index, H is the terrain height of the area to be evaluated, α is its preset proportional coefficient, C is the terrain slope of the area to be evaluated, β is its preset proportional coefficient, S is the predicted disaster area, T is the predicted duration, and ρ is (S 2 +T 3 ), C1 is the preset proportional coefficient, C2 is the constant correction index, and α, β, and ρ are all greater than zero.

[0044] Furthermore, the comprehensive weather index QS is compared with the preset assessment threshold, and the risk level is divided according to the comparison result, which specifically includes: setting a series of thresholds, which correspond to different gale disaster risk levels, comparing the obtained comprehensive weather index with the preset threshold, and dividing the risk level according to the comparison result:

[0045] When QS<30, the forecast is low risk and the public can carry out outdoor activities normally, but they need to pay attention to weather changes;

[0046] When 30≤QS<60, the forecast is medium risk, and the public is advised to reduce unnecessary outdoor activities and secure loose items to prevent them from being blown away by the wind;

[0047] When 60≤QS<80, the forecast is high risk, and the public is advised to avoid outdoor activities. It is recommended that relevant departments strengthen patrols to ensure the stability of public safety facilities. Residents should ensure that doors and windows are closed and outdoor items are secured;

[0048] When QS ≥ 80, the forecast is extremely high risk. The public is advised to stay indoors and avoid all outdoor activities. Relevant departments should initiate emergency response measures, such as evacuating residents in high-risk areas and closing public facilities that may be affected.

[0049] Among them, QS is the comprehensive weather index.

[0050] The present invention further provides a method for predicting a gale disaster weather index based on deep learning. The method for predicting a gale disaster weather index is performed by the gale disaster weather index prediction system described above, and the specific steps include:

[0051] Step 1: Collect the affected area and duration of the strong wind disaster events that occurred in the area to be evaluated, and collect the historical meteorological data of the strong wind disaster events in the area to be evaluated one week ago, the meteorological data includes wind speed, humidity, temperature, air pressure, precipitation and cloud cover;

[0052] Step 2: Establish an index prediction model based on the long short-term memory network, input the historical meteorological data into the index prediction model as a training set, use the corresponding disaster area and duration of the gale disaster as labels, train the model, and obtain a trained index prediction model;

[0053] Step 3: Collect meteorological data of the area to be evaluated in the past week and input it into the trained index prediction model to obtain the predicted disaster area and predicted duration of the strong wind disaster weather. Use the rasterization method to collect terrain height and slope data of the area to be evaluated;

[0054] Step 4: Use the collected terrain height and slope data of the area to be evaluated to correct the predicted disaster-stricken area and predicted duration of the strong wind disaster weather to obtain a comprehensive weather index, compare the comprehensive weather index with the preset assessment threshold, and divide the risk level according to the comparison results.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] The present invention effectively solves the limitations of traditional prediction methods in processing complex nonlinear relationships and time series data by introducing a long short-term memory network model based on deep learning. The LSTM model can learn the deep-level association between factors such as wind speed, humidity, temperature and gale disaster events from historical meteorological data, thereby improving the accuracy of prediction. At the same time, terrain factors have a significant impact on wind speed and wind direction, which may change the path and impact range of gale, and are therefore key variables for accurate prediction. The terrain height and slope data of the area to be evaluated are collected using a rasterization method. Through the correction of terrain data, the regulatory effect of terrain on the impact of gale disasters is fully considered, making the prediction results more practical and closer to the actual disaster situation. This prediction method that comprehensively considers meteorological factors and terrain characteristics not only improves the accuracy and reliability of gale disaster prediction, but also can provide more scientific and effective decision-making support for disaster prevention and emergency management, significantly improving the efficiency and effectiveness of disaster response. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a schematic diagram of the overall system module of the present invention;

[0058] Figure 2 It is a schematic diagram of the overall method flow of the present invention. DETAILED DESCRIPTION

[0059] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0060] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0061] Example:

[0062] See also Figure 1 The present invention provides a gale disaster weather index prediction system based on deep learning, comprising:

[0063] A data collection module is used to collect the affected area and duration of the strong wind disaster events that occurred in the history of the area to be evaluated, and collect the historical meteorological data of the strong wind disaster events in the area to be evaluated, which is traced back to one week ago. The meteorological data includes wind speed, humidity, temperature, air pressure, precipitation and cloud cover;

[0064] In this embodiment, the specific logic for collecting the wind speed, humidity, temperature and air pressure of the area to be evaluated one week before the occurrence of a strong wind disaster event in history is:

[0065] The weather data recorded in the history of the Meteorological Bureau is used to obtain the wind speed, humidity, temperature and air pressure in the week before the strong wind disaster event occurred in the area to be assessed;

[0066] Select n measurement points around the area to be evaluated, covering the four directions of east, west, south and north, and select measurement points in equal quantities and distances in each direction;

[0067] The specific logic for collecting meteorological data for the area to be evaluated in the past week is as follows:

[0068] Install anemometers, hygrometers, thermometers and barometers at the measurement points;

[0069] Record wind speed data, humidity data, temperature data and air pressure data at different measurement points to obtain the average wind speed, average humidity, average temperature and average air pressure of the area to be evaluated;

[0070] The formula for obtaining the average wind speed is:

[0071]

[0072] in, is the average wind speed, v i represents the wind speed measured at the i-th measurement point;

[0073] The formula used to obtain the average humidity is:

[0074]

[0075] in, is the average humidity, RH i represents the humidity measured at the i-th measurement point;

[0076] The formula used to obtain the average temperature is:

[0077]

[0078] in, is the average temperature, Temp i represents the temperature measured at the i-th measurement point;

[0079] The formula for obtaining the average air pressure is:

[0080]

[0081] in, is the average air pressure, PRES i represents the air pressure measured at the i-th measurement point;

[0082] The specific logic of using the rasterization method to collect terrain height and slope data of the area to be evaluated is:

[0083] The area to be evaluated is divided into regular small grids according to its size, and then the average height and slope are calculated in each grid unit to generate a continuous terrain surface model. Each unit represents an average value, and the heights and slopes of all grid units are weighted averaged again. The final values ​​are used as the terrain height H and slope C of the area to be evaluated.

[0084] The specific logic for collecting the precipitation M and cloud cover N in the week before the strong wind disaster event occurred in the history of the area to be assessed is:

[0085] The historical weather data recorded by the Meteorological Bureau are used to obtain the daily average precipitation M and daily average cloud cover N in the week before the strong wind disaster event in the area to be assessed;

[0086] The collected cloud amount N is the proportion of the sky in the disaster area covered by clouds when a strong wind disaster occurs, expressed as a percentage, ranging from 0% to 100%;

[0087] The specific logic for collecting precipitation and cloud cover data for the area to be evaluated in the past week is as follows:

[0088] The precipitation and cloud cover in the area to be evaluated in the past week are collected through the historical weather data recorded by the Meteorological Bureau and the forecast results of the weather forecast;

[0089] The specific method for collecting the disaster-affected area S is:

[0090] Use satellite imagery to identify and measure the affected areas, and use image processing software to analyze the images, identify the affected areas and calculate their size;

[0091] The specific method for collecting the disaster duration T is:

[0092] Use remote sensing technology to monitor the development of disaster events. By analyzing continuous remote sensing images, the duration T of the disaster event can be determined.

[0093] Historical meteorological data is a key input for training forecasting models. By analyzing patterns and trends in historical data, the model can learn the laws of weather changes and thus improve the accuracy of predicting future weather.

[0094] The model building module builds an index prediction model based on the long short-term memory network, inputs historical meteorological data as a training set into the index prediction model, uses the corresponding disaster area and duration of the gale disaster as labels, trains the model, and obtains a trained index prediction model;

[0095] The process of establishing an index prediction model based on a long short-term memory network specifically includes:

[0096] The historical meteorological data are structured into multiple groups of time series data according to their types and used as training sets. The affected area and duration of the gale disaster are used as labels. A deep learning model based on a long short-term memory network is constructed. The training set and the labels are input into the deep learning model, and the model is trained to obtain a trained index prediction model. The historical meteorological data include wind speed, humidity, temperature, air pressure, precipitation and cloud cover in the area to be evaluated.

[0097] LSTM is a special recurrent neural network (RNN) that can effectively process and predict long-term dependencies in time series. In meteorological data, weather patterns often have complex time dependencies, and LSTM can capture these patterns, thereby improving the accuracy of predictions; LSTM models can directly map from raw input data (wind speed, temperature, humidity, etc.) to output labels (affected area and duration) to achieve end-to-end learning. This direct mapping simplifies the model training process and may improve the accuracy of predictions.

[0098] The prediction module collects the meteorological data of the area to be evaluated in the past week and inputs it into the trained index prediction model to obtain the predicted disaster area and duration of the strong wind disaster weather. The terrain height and slope data of the area to be evaluated are collected using the rasterization method.

[0099] In this embodiment, the specific logic for obtaining the predicted disaster-affected area S and the predicted duration T of the strong wind disaster weather is as follows:

[0100] Collect meteorological data of the area to be evaluated in the past week, including wind speed, humidity, temperature, air pressure, precipitation and cloud cover in the area to be evaluated, input them into the trained index prediction model, and output the predicted disaster-affected area S and predicted duration T of the strong wind disaster weather;

[0101] Real-time collection of meteorological data can ensure that the model inputs the latest information, which helps improve the real-time and accuracy of the forecast. Compared with technologies that rely on historical data, real-time data can better reflect current weather conditions; combined with terrain height and slope data, it can more comprehensively assess the impact of high wind disasters on different terrain areas. This multi-dimensional data fusion helps improve the model's predictive ability because it takes into account the impact of terrain on wind speed and wind direction, which can help the model better adapt to the forecasting needs in different geographical environments, thereby improving the model's generalization ability.

[0102] The comprehensive assessment module is used to use the collected terrain height and slope data of the area to be assessed to correct the predicted disaster area and predicted duration of the gale disaster weather, obtain a comprehensive weather index, compare the comprehensive weather index with a preset assessment threshold, and classify the risk level according to the comparison result;

[0103] In this embodiment, the predicted disaster area and predicted duration of gale weather are corrected by using the collected terrain height and slope data of the area to be evaluated, and the formula for obtaining the comprehensive weather index is as follows:

[0104]

[0105] Among them, QS is the comprehensive weather index, H is the terrain height of the area to be evaluated, α is its preset proportional coefficient, C is the terrain slope of the area to be evaluated, β is its preset proportional coefficient, S is the predicted disaster area, T is the predicted duration, and ρ is (S 2 +T 3 ), C1 is a constant correction index, α, β, ρ are all greater than zero; when the terrain height H of the area to be evaluated increases, the comprehensive weather index QS increases, because it will cause a change in wind direction, making the wind unevenly distributed in different terrain areas. This change in wind direction may cause some areas to be affected by stronger winds, thereby increasing the risk of disasters; when the terrain slope C increases, the comprehensive weather index QS increases, and steep slopes may cause stronger local wind effects, such as slope winds or downhill winds. These local wind effects may produce extreme wind speeds in specific areas, increasing the risk of disasters; when the predicted disaster-affected area S increases, QS increases; when the predicted duration T increases, QS will also increase accordingly; that is, H, C, S, T and the comprehensive weather index QS are positively correlated;

[0106] The process of comparing the comprehensive weather index QS with the preset assessment threshold and classifying the risk level according to the comparison result specifically includes: setting a series of thresholds corresponding to different gale disaster risk levels, comparing the obtained comprehensive weather index with the preset threshold, and classifying the risk level according to the comparison result:

[0107] When QS<30, the forecast is low risk and the public can carry out outdoor activities normally, but they need to pay attention to weather changes;

[0108] When 30≤QS<60, the forecast is medium risk, and the public is advised to reduce unnecessary outdoor activities and secure loose items to prevent them from being blown away by the wind;

[0109] When 60≤QS<80, the forecast is high risk, and the public is advised to avoid outdoor activities. It is recommended that relevant departments strengthen patrols to ensure the stability of public safety facilities. Residents should ensure that doors and windows are closed and outdoor items are secured;

[0110] When QS ≥ 80, the forecast is extremely high risk. The public is advised to stay indoors and avoid all outdoor activities. Relevant departments should initiate emergency response measures, such as evacuating residents in high-risk areas and closing public facilities that may be affected.

[0111] Among them, QS is the comprehensive weather index.

[0112] See also Figure 2 The present invention also provides a method for predicting a gale disaster weather index based on deep learning. The method for predicting a gale disaster weather index is performed by the gale disaster weather index prediction system described above, and the specific steps include:

[0113] Step 1: Collect the affected area and duration of the strong wind disaster events that occurred in the area to be evaluated, and collect the historical meteorological data of the strong wind disaster events in the area to be evaluated one week ago, the meteorological data includes wind speed, humidity, temperature, air pressure, precipitation and cloud cover;

[0114] Step 2: Establish an index prediction model based on the long short-term memory network, input the historical meteorological data into the index prediction model as a training set, use the corresponding disaster area and duration of the gale disaster as labels, train the model, and obtain a trained index prediction model;

[0115] Step 3: Collect meteorological data of the area to be evaluated in the past week and input it into the trained index prediction model to obtain the predicted disaster area and predicted duration of the strong wind disaster weather. Use the rasterization method to collect terrain height and slope data of the area to be evaluated;

[0116] Step 4: Use the collected terrain height and slope data of the area to be evaluated to correct the predicted disaster-stricken area and predicted duration of the strong wind disaster weather to obtain a comprehensive weather index, compare the comprehensive weather index with the preset assessment threshold, and divide the risk level according to the comparison results.

[0117] The specific values ​​of α, β and ρ in the formula are generally determined by technical personnel in this field according to actual conditions. The technical personnel in this field collect multiple groups of sample data, and set corresponding preset proportional coefficients for each group of sample data. The preset proportional coefficients and the collected sample data are substituted into the formula. Through repeated experiments and parameter adjustments, the accuracy of the model output and the rationality of the results are observed, and these factor coefficients are gradually adjusted. The performance and effect of the model under different parameter settings are compared to find the optimal coefficient combination, and the calculated factor coefficients are screened and averaged to obtain the values ​​of α, β and ρ.

[0118] In addition, the size of the preset factor coefficient is to quantify each parameter to obtain a specific value. In order to facilitate subsequent comparison, the size of the coefficient depends on the amount of sample data and the preset proportional coefficient initially set by technical personnel in this field for each set of sample data. It is not unique as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0119] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0120] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0121] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0122] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A high wind disaster weather index prediction system based on deep learning, characterized in that: Specifically include: A data collection module is used to collect the affected area and duration of the strong wind disaster events that occurred in the history of the area to be evaluated, and collect the historical meteorological data of the strong wind disaster events in the area to be evaluated, which is traced back to one week ago. The meteorological data includes wind speed, humidity, temperature, air pressure, precipitation and cloud cover; The model building module builds an index prediction model based on the long short-term memory network, inputs historical meteorological data as a training set into the index prediction model, uses the corresponding disaster area and duration of the gale disaster as labels, trains the model, and obtains a trained index prediction model; The prediction module collects the meteorological data of the area to be evaluated in the past week and inputs it into the trained index prediction model to obtain the predicted disaster area and duration of the strong wind disaster weather. The terrain height and slope data of the area to be evaluated are collected using the rasterization method. The comprehensive assessment module is used to use the collected terrain height and slope data of the area to be assessed to correct the predicted disaster-stricken area and predicted duration of strong wind disaster weather, obtain a comprehensive weather index, compare the comprehensive weather index with a preset assessment threshold, and divide the risk level according to the comparison results.

2. The high wind disaster weather index prediction system based on deep learning according to claim 1 is characterized by: The specific logic for collecting the wind speed, humidity, temperature and air pressure in the area to be assessed one week before the occurrence of a strong wind disaster event in history is as follows: The weather data recorded in the history of the Meteorological Bureau is used to obtain the wind speed, humidity, temperature and air pressure in the week before the strong wind disaster event occurred in the area to be assessed; Select n measurement points around the area to be evaluated, covering the four directions of east, west, south and north, and select measurement points in equal quantities and distances in each direction; The specific logic for collecting meteorological data for the area to be evaluated in the past week is as follows: Install anemometers, hygrometers, thermometers and barometers at the measurement points; Record wind speed data, humidity data, temperature data and air pressure data at different measurement points to obtain the average wind speed, average humidity, average temperature and average air pressure of the area to be evaluated; The formula for obtaining the average wind speed is: in, is the average wind speed, v i represents the wind speed measured at the i-th measurement point; The formula for obtaining the average humidity is: in, is the average humidity, RH i represents the humidity measured at the i-th measurement point; The formula used to obtain the average temperature is: in, is the average temperature, Temp i represents the temperature measured at the i-th measurement point; The formula for obtaining the average air pressure is: in, is the average air pressure, PRES i represents the air pressure measured at the i-th measurement point.

3. The high wind disaster weather index prediction system based on deep learning according to claim 1 is characterized by: The specific logic of using the rasterization method to collect the terrain height H and slope data C of the area to be evaluated is: The area to be evaluated is divided into regular small grids according to its size, and then the average height and slope are calculated in each grid unit to generate a continuous terrain surface model. Each unit represents an average value, and the heights and slopes of all grid units are weighted averaged again. The final values ​​are used as the terrain height H and slope C of the area to be evaluated.

4. The high wind disaster weather index prediction system based on deep learning according to claim 2 is characterized by: The specific logic for collecting the precipitation M and cloud cover N in the week before the strong wind disaster event occurred in the history of the area to be assessed is: The historical weather data recorded by the Meteorological Bureau are used to obtain the daily average precipitation M and daily average cloud cover N in the week before the strong wind disaster event in the area to be assessed; The collected cloud amount N is the proportion of the sky in the disaster area covered by clouds when a strong wind disaster occurs, expressed as a percentage, ranging from 0% to 100%; The specific logic for collecting precipitation and cloud cover data for the area to be evaluated in the past week is as follows: The precipitation and cloud cover in the area to be evaluated in the past week are collected through the historical weather data recorded by the Meteorological Bureau and the prediction results of the weather forecast.

5. The high wind disaster weather index prediction system based on deep learning according to claim 1 is characterized by: The process of establishing an index prediction model based on a long short-term memory network specifically includes: Historical meteorological data are organized into multiple groups of time series data according to their types and used as training sets. The affected area and duration of the strong wind disasters are used as labels. A deep learning model based on the long short-term memory network is constructed. The training set and labels are input into the deep learning model, and the model is trained to obtain a trained index prediction model. The historical meteorological data include wind speed, humidity, temperature, air pressure, precipitation and cloud cover in the area to be evaluated.

6. The high wind disaster weather index prediction system based on deep learning according to claim 1 is characterized by: The specific logic for obtaining the predicted disaster-affected area S and predicted duration T of high wind disaster weather is as follows: The meteorological data of the area to be evaluated in the past week are collected, including wind speed, humidity, temperature, air pressure, precipitation and cloud cover in the area to be evaluated, and input into the trained exponential prediction model to output the predicted disaster-stricken area S and predicted duration T of the strong wind disaster weather.

7. The high wind disaster weather index prediction system based on deep learning according to claim 6 is characterized by: The predicted disaster area and predicted duration of strong wind weather are corrected by using the terrain height and slope data collected in the area to be evaluated. The formula for the comprehensive weather index is as follows: Among them, QS is the comprehensive weather index, H is the terrain height of the area to be evaluated, α is its preset proportional coefficient, C is the terrain slope of the area to be evaluated, β is its preset proportional coefficient, S is the predicted disaster area, T is the predicted duration, and ρ is (S 2 +T 3 ), C1 is the preset proportional coefficient, C2 is the constant correction index, and α, β, and ρ are all greater than zero.

8. The high wind disaster weather index prediction system based on deep learning according to claim 7 is characterized by: The process of comparing the comprehensive weather index QS with the preset assessment threshold and classifying the risk level according to the comparison result specifically includes: setting a series of thresholds corresponding to different gale disaster risk levels, comparing the obtained comprehensive weather index with the preset threshold, and classifying the risk level according to the comparison result: When QS<30, the forecast is low risk and the public can carry out outdoor activities normally, but they need to pay attention to weather changes; When 30≤QS<60, the forecast is medium risk, and the public is advised to reduce unnecessary outdoor activities and secure loose items to prevent them from being blown away by the wind; When 60≤QS<80, the forecast is high risk, and the public is advised to avoid outdoor activities. It is recommended that relevant departments strengthen patrols to ensure the stability of public safety facilities. Residents should ensure that doors and windows are closed and outdoor items are secured; When QS ≥ 80, the forecast is extremely high risk, and the public is advised to stay indoors and avoid all outdoor activities. Relevant departments should initiate emergency response measures, such as evacuating residents in high-risk areas and closing public facilities that may be affected; Among them, QS is the comprehensive weather index.

9. A method for predicting gale disaster weather index based on deep learning, characterized by: The method for predicting a gale disaster weather index is obtained by using a gale disaster weather index prediction system based on deep learning as described in any one of claims 1 to 8, and comprises: Step 1: Collect the affected area and duration of the strong wind disaster events that occurred in the area to be evaluated, and collect the historical meteorological data of the strong wind disaster events in the area to be evaluated one week ago, the meteorological data includes wind speed, humidity, temperature, air pressure, precipitation and cloud cover; Step 2: Establish an index prediction model based on the long short-term memory network, input historical meteorological data into the index prediction model as a training set, use the corresponding disaster area and duration of the gale disaster as labels, train the model, and obtain a trained index prediction model; Step 3: Collect meteorological data of the area to be evaluated in the past week and input it into the trained index prediction model to obtain the predicted disaster area and predicted duration of the strong wind disaster weather. Use the rasterization method to collect terrain height and slope data of the area to be evaluated; Step 4: Use the collected terrain height and slope data of the area to be evaluated to correct the predicted disaster-stricken area and predicted duration of the strong wind disaster weather to obtain a comprehensive weather index, compare the comprehensive weather index with the preset assessment threshold, and divide the risk level according to the comparison results.

Images