Meteorological element forecasting method based on artificial intelligence model

Abstract

The invention discloses a meteorological element forecasting method based on an artificial intelligence model, and the method comprises the following steps: firstly, collecting meteorological data from a plurality of data sources of a ground meteorological station, a sounding station, a meteorological satellite, a weather radar, and a microwave radiometer, the data comprising key elements such as temperature, humidity, wind speed, wind direction, air pressure, precipitation, etc. According to the system, data cleaning and conversion technologies are adopted, abnormal values are removed, missing data are filled up, data quality is ensured, original data are converted into feature vectors suitable for a machine learning algorithm, and finally a meteorological element forecasting product with high temporal-spatial resolution is obtained through calculation based on the machine learning algorithm. The CNN model is adopted in the learning model, the CNN reduces the number of parameters of the model through a local connection and weight sharing mechanism, the calculation cost is reduced, the generalization ability of the model is improved, high-resolution, high-frequency and long-term trend temperature prediction can be achieved, and more accurate and more reliable weather forecast is provided for people.

Description

Technical Field

[0001] The present invention relates to the technical field of weather forecasting, and in particular to a weather element forecasting method based on an artificial intelligence model. Background Art

[0002] In recent years, global meteorological disasters have been on the rise. If we can obtain a longer-term forecast of meteorological elements, we can help with disaster prevention and mitigation more accurately.

[0003] The prior art discloses an extended-term meteorological element forecasting method and system based on an artificial intelligence model with announcement number: CN115657163B, including: collecting real-time information of meteorological elements and 10-30 days model forecast information including 71 members; interpolating the model forecast information to the grid points in the Yangtze River Delta region; preferentially screening the members in the model forecast information to determine the preferred members of the model forecast information; training the preferred members using the artificial intelligence model to determine the optimal parameter matching scheme of the training model and the weight values ​​of the calculated preferred members; weighted integration of the preferred members of the model forecast information according to the weight values, and outputting the forecast results. The device introduces a machine learning model to the 10-30 days extended-term forecast, providing a new idea for solving the difficult problems of meteorological forecasting; the forecast results of the extended-term forecast method proposed in the present invention are compared with the mainstream model ensemble forecast results, which improves the accuracy to a certain extent and achieves better results.

[0004] The extended-term meteorological element forecasting method of the above-mentioned artificial intelligence model discloses an extended-term meteorological element forecasting method and system based on the artificial intelligence model, introduces the currently widely used machine learning model into the extended-term forecast, and provides a new idea for solving the difficult problems of meteorological forecasting. The above-mentioned learning model does not adopt the CNN model, and cannot reduce the number of model parameters through the mechanism of local connection and weight sharing. The calculation cost is high and the generalization ability of the model is low. Therefore, it is necessary to design a meteorological element forecasting method based on the artificial intelligence model with strong practicality. Summary of the invention

[0005] The purpose of the present invention is to provide a meteorological element forecasting method based on an artificial intelligence model to solve the problems raised in the above-mentioned background technology.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a meteorological element forecasting method based on an artificial intelligence model, comprising the following steps: S1. Meteorological data collection Meteorological data are collected from multiple data sources such as ground meteorological stations, sounding stations, meteorological satellites, weather radars, and microwave radiometers. These data include key elements such as temperature, humidity, wind speed, wind direction, air pressure, and precipitation. The system uses data cleaning and conversion technology to remove outliers and fill in missing data to ensure data quality, and convert raw data into feature vectors suitable for machine learning algorithms.

[0007] S2. Learning model application A variety of deep learning models are used, including convolutional neural networks (CNN), recurrent neural networks (RNN) and generative adversarial networks (GAN) to capture complex regularities and patterns in meteorological data.

[0008] S2.1 precipitation forecast uses CNN to perform image recognition on satellite cloud images and weather radar data, combined with RNN to process time series data such as ground observations, sounding observations and microwave radiometers, to achieve high-precision prediction of future precipitation intensity and distribution.

[0009] S2.2 Temperature forecast: Through the RNN variant of LSTM, historical temperature data is learned, and factors such as seasonal changes and geographical location are considered to predict future temperature change trends.

[0010] S2.3 Wind forecast: Combining CNN’s feature extraction of ground observation data, high-altitude observation data and radar data with RNN’s modeling capabilities for time series, it predicts wind speed and direction, providing accurate support for wind power generation, air transportation, etc.

[0011] According to the above technical solution, the precipitation forecast in step S2.1 first relies on the collection of multi-source data, including ground meteorological stations, sounding meteorological stations, meteorological satellites, weather radars, microwave radiometers and other data sources. These data cover multiple dimensions such as precipitation amount, precipitation intensity, and precipitation type. The artificial intelligence model will preprocess these raw data, including data cleaning, which can remove outliers and fill in missing data, data conversion, which can convert the raw data into a format recognizable by the model, and feature extraction, including extracting key features useful for precipitation forecasting.

[0012] According to the above technical solution, the LSTM in step S2.2 refers to a long short-term memory network. Temperature forecasting relies on data from multiple sources, including ground meteorological stations, satellite observations, and high-altitude detection. These data provide rich information about current and past temperature conditions. The artificial intelligence model will clean, transform, and extract features from these raw data to eliminate noise, fill in missing values, and extract key features that are useful for temperature prediction.

[0013] According to the above technical solution, in step S2.2, for temperature forecasting, deep learning models of convolutional neural networks and recurrent neural networks will be utilized, and the models can capture the spatiotemporal characteristics in temperature data, such as seasonal changes and diurnal changes.

[0014] According to the above technical solution, the CNN model is adopted in S2. CNN reduces the number of model parameters, reduces the computational cost, and improves the generalization ability of the model through the mechanism of local connection and weight sharing.

[0015] According to the above technical solution, the convolutional neural network model can capture the spatiotemporal characteristics of wind data, such as wind propagation path and intensity changes. The meteorological data is read and entered into the model, and then the artificial intelligence model is used to clean, convert and extract features from the raw data to eliminate noise, fill in missing values, and extract key features that are useful for wind forecasting, including pressure gradients and terrain effects.

[0016] According to the above technical solution, CNN is usually composed of multiple layers such as convolutional layers, pooling layers and fully connected layers. These layers are stacked to extract high-level features in the image. The convolutional layer performs sliding window calculations on the input image through convolution kernel filters to extract local features in the image, and then pooling operations can be performed. The pooling layer further reduces the size and number of parameters of the feature map by downsampling while retaining important information.

[0017] Compared with the prior art, the beneficial effect achieved by the present invention is that the learning model in the present invention adopts the CNN model, which reduces the number of model parameters, reduces the computational cost, and improves the generalization ability of the model through the mechanism of local connection and weight sharing.

[0018] By setting up a deep learning model with convolutional neural networks and recurrent neural networks, the model can capture the spatiotemporal characteristics of temperature data, such as seasonal changes and daily changes, and show a high degree of accuracy and meticulousness in temperature forecasting. Through the application of technologies such as multi-source data fusion, deep learning model construction and training, and physical mechanism fusion, it can achieve high-resolution, high-frequency and long-term trend temperature forecasts, providing people with more accurate and reliable weather forecasts; By setting up a long short-term memory network learning module, temperature forecasts rely on data from multiple sources, including ground-based weather stations, satellite observations, and high-altitude detection, which provide rich information about current and past temperature conditions. The artificial intelligence model will clean, transform, and extract features from these raw data to eliminate noise, fill in missing values, and extract key features that are useful for temperature prediction. DETAILED DESCRIPTION

[0019] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0020] The present invention provides a technical solution: a meteorological element forecasting method based on an artificial intelligence model, comprising the following steps: S1. Meteorological data collection Meteorological data are collected from multiple data sources such as ground meteorological stations, sounding stations, meteorological satellites, weather radars, and microwave radiometers. These data include key elements such as temperature, humidity, wind speed, wind direction, air pressure, and precipitation. The system uses data cleaning and conversion technology to remove outliers and fill in missing data to ensure data quality, and convert raw data into feature vectors suitable for machine learning algorithms.

[0021] S2. Reference Learning Model A variety of deep learning models are used, including convolutional neural network (CNN), recurrent neural network (RNN) and generative adversarial network (GAN) to capture the complex laws and patterns in meteorological data. The CNN model is adopted. CNN reduces the number of model parameters and computational costs through the mechanism of local connection and weight sharing, and improves the generalization ability of the model.

[0022] S2.1 Precipitation forecast, using CNN to perform image recognition on satellite cloud images and weather radar, combined with RNN to process time series data such as ground observations, sounding observations and microwave radiometers, to achieve high-precision prediction of future precipitation intensity and distribution. CNN is usually composed of multiple layers such as convolutional layers, pooling layers and fully connected layers. These layers are stacked to extract high-level features in the image. The convolutional layer performs sliding window calculations on the input image through convolution kernel filters to extract local features in the image, and then pooling operations can be performed. The pooling layer further reduces the size and number of parameters of the feature map by downsampling, while retaining important information. Precipitation forecasting first relies on the collection of multi-source data, including ground meteorological stations, sounding meteorological stations, meteorological satellites, weather radars, microwave radiometers and other data sources. These data cover multiple dimensions such as precipitation, precipitation intensity, and precipitation type. The artificial intelligence model will preprocess these raw data, including data cleaning, which can remove outliers and fill in missing data, data conversion, which can convert raw data into a format recognizable by the model, and feature extraction, including extracting key features useful for precipitation forecasting.

[0023] S2.2 Temperature forecast: Through the RNN variant of LSTM, historical temperature data is learned, and factors such as seasonal changes and geographical location are considered to predict future temperature trends. LSTM refers to the long short-term memory network. Temperature forecasting relies on data from multiple sources, including ground weather stations, satellite observations, and high-altitude detection. These data provide rich information about current and past temperature conditions. The artificial intelligence model will clean, transform and extract features from these raw data to eliminate noise, fill in missing values, and extract key features that are useful for temperature prediction. For temperature forecasting, deep learning models of convolutional neural networks and recurrent neural networks will be used. The model can capture the spatiotemporal characteristics of temperature data, such as seasonal changes and daily changes.

[0024] S2.3 Wind forecast: Combining CNN's feature extraction of ground observation data, high-altitude observation data and radar data with RNN's modeling capabilities for time series, wind speed and direction are predicted to provide accurate support for wind power generation, air transportation, etc. The convolutional neural network model CNN can capture the spatiotemporal characteristics of wind data, such as wind propagation path and intensity changes. By reading meteorological data and entering it into the model, the artificial intelligence model will then clean, convert and extract features from the raw data to eliminate noise, fill in missing values, and extract key features that are useful for wind forecasting, including pressure gradients and terrain effects.

[0025] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or apparatus that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or apparatus.

[0026] Finally, it should be noted that the above are only preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A meteorological element forecasting method based on an artificial intelligence model, characterized in that: The following steps are involved: S1. Meteorological data collection Meteorological data are collected from multiple data sources such as ground meteorological stations, sounding meteorological stations, meteorological satellites, weather radars, microwave radiometers, etc. These data include key elements such as temperature, humidity, wind speed, wind direction, air pressure, precipitation, etc. The system uses data cleaning and conversion technology to remove outliers and fill in missing data to ensure data quality, and convert raw data into feature vectors suitable for machine learning algorithms; S2. Learning model application Use a variety of deep learning models, including convolutional neural networks (CNN), recurrent neural networks (RNN), and generative adversarial networks (GAN) to capture complex regularities and patterns in meteorological data; S2.1 precipitation forecast, using CNN to perform image recognition on satellite cloud images and weather radar, combined with RNN to process time series data such as ground observations, sounding observations and microwave radiometers, to achieve high-precision prediction of future precipitation intensity and distribution; S2.2 Temperature forecast: Through the RNN variant of LSTM, historical temperature data is learned, and factors such as seasonal changes and geographical location are considered to predict future temperature change trends; S2.3 Wind forecast: Combining CNN’s feature extraction of ground observation data, high-altitude observation data and radar data with RNN’s modeling capabilities for time series, it predicts wind speed and direction, providing accurate support for wind power generation, air transportation, etc.

2. The meteorological element forecasting method based on the artificial intelligence model according to claim 1, characterized in that: The precipitation forecast in step S2.1 first relies on the collection of multi-source data, including ground meteorological stations, sounding meteorological stations, meteorological satellites, weather radars, microwave radiometers and other data sources. These data cover multiple dimensions such as precipitation, precipitation intensity, and precipitation type. The artificial intelligence model will preprocess these raw data, including data cleaning, which can remove outliers and fill in missing data, data conversion, which can convert the raw data into a format recognizable by the model, and feature extraction, including extracting key features useful for precipitation forecasting.

3. The meteorological element forecasting method based on the artificial intelligence model according to claim 2 is characterized in that: The LSTM in step S2.2 refers to the long short-term memory network. Temperature forecasting relies on data from multiple sources, including ground weather stations, satellite observations, and high-altitude detection. These data provide rich information about current and past temperature conditions. The artificial intelligence model will clean, transform, and extract features from these raw data to eliminate noise, fill in missing values, and extract key features that are useful for temperature prediction.

4. The meteorological element forecasting method based on the artificial intelligence model according to claim 3 is characterized in that: In step S2.2, for temperature forecasting, deep learning models of convolutional neural networks and recurrent neural networks are used. The models can capture the spatiotemporal characteristics of temperature data, such as seasonal changes and diurnal changes.

5. The meteorological element forecasting method based on the artificial intelligence model according to claim 4 is characterized in that: The S2 described above adopts the CNN model. Through the mechanism of local connection and weight sharing, CNN reduces the number of model parameters, reduces the computational cost, and improves the generalization ability of the model.

6. The meteorological element forecasting method based on the artificial intelligence model according to claim 5 is characterized in that: The convolutional neural network model can capture the spatiotemporal characteristics of wind data, such as wind propagation paths and intensity changes. By reading meteorological data and entering it into the model, the artificial intelligence model is then used to clean, convert and extract features from the raw data to eliminate noise, fill in missing values, and extract key features that are useful for wind forecasting, including pressure gradients and terrain effects.

7. The meteorological element forecasting method based on the artificial intelligence model according to claim 6 is characterized in that: CNN is usually composed of multiple layers such as convolutional layers, pooling layers and fully connected layers. These layers are stacked to extract high-level features in the image. The convolutional layer performs sliding window calculations on the input image through convolution kernel filters to extract local features in the image, and then pooling operations can be performed. The pooling layer further reduces the size and number of parameters of the feature map by downsampling while retaining important information.