CN-121980146-A - Airport visibility prediction method, storage medium and related device

CN121980146ACN 121980146 ACN121980146 ACN 121980146ACN-121980146-A

Abstract

The invention relates to the technical field of airport meteorological element prediction, and discloses a prediction method of airport visibility, a storage medium and a related device. The method comprises the steps of collecting numerical weather forecast data and weather live data of an airport, which take effect in an airport and a nearby area, extracting data related to visibility from the numerical weather forecast data and the weather live data respectively to serve as numerical weather element forecast data and weather element live data, carrying out two-dimensional extraction on grid type data in the numerical weather element forecast data and the weather element live data to obtain first space related characteristics, carrying out one-dimensional extraction on single-point time sequence weather data in the numerical weather element forecast data and the weather element live data to obtain first time evolution characteristics, and inputting the first space related characteristics and the first time evolution characteristics into a pre-trained visibility forecast model to obtain a short-term prediction result and an approaching prediction result of the visibility. The method can improve accuracy and efficiency of visibility prediction.

Inventors

LI RONGPEI
WANG JINGYAO
LUO MINGJIAN
HU SHUJUN
SHEN WUYANG
YU PENG
LIU JIE
ZHOU MINGJIA
BAI HANWEN
WU RUIHAO

Assignees

中国民用航空珠海空中交通管理站

Dates

Publication Date: 20260505
Application Date: 20260121

Claims (10)

1. A method for predicting airport visibility, comprising: Acquiring numerical weather forecast data and weather live data of an airport, which take effect in the airport and a nearby area; the method comprises the steps of respectively extracting data related to visibility from numerical weather forecast data and weather live data as numerical weather element forecast data and weather element live data; Two-dimensional extraction is carried out on grid type data in the digital weather element forecast data and the weather element live data to obtain first space correlation characteristics, and one-dimensional extraction is adopted on single-point time sequence weather data in the digital weather element forecast data and the weather element live data to obtain first time evolution characteristics; And inputting the first spatial correlation characteristic and the first time evolution characteristic into a pre-trained visibility prediction model to obtain a short-term visibility prediction result and an adjacent prediction result.
2. The method for predicting airport visibility according to claim 1, wherein the step of inputting the spatial correlation feature and the time evolution feature into a pre-trained visibility prediction model to obtain a short-term visibility prediction result and an adjacent visibility prediction result, wherein the visibility prediction model is trained by the steps of: Collecting global weather analysis data and airport historical weather live data for the last several years; The method comprises the steps of extracting data related to visibility of airports and nearby areas from global weather analysis data as weather element analysis data, and extracting data related to visibility from airport historical weather live data as airport weather element historical live data; two-dimensional extraction is carried out on grid type data in the meteorological element analysis data and the airport meteorological element historical live data to obtain second space correlation characteristics, and one-dimensional extraction is adopted on single-point time sequence meteorological data in the meteorological element analysis data and the airport meteorological element historical live data to obtain second time evolution characteristics; Converting the second spatial correlation feature and the second time evolution feature into predictive initial model training data, taking the predictive initial model training data as input of a decision tree model, taking airport historical visibility in airport meteorological element historical live data as output of the decision tree model, and training the decision tree model to obtain a visibility prediction model.
3. The method of predicting airport visibility of claim 2, wherein said step of converting the second spatial correlation feature and the second time evolution feature into predicted initial model training data comprises: converting the data of the second spatial correlation feature and the second time evolution feature into a unified format; Removing abnormal values in the second spatial correlation feature and the second time evolution feature by an abnormal analysis method; and supplementing missing values in the second spatial correlation feature and the second time evolution feature through a difference algorithm.
4. The method for predicting airport visibility of claim 2, wherein the decision tree model is a gradient-lifting decision tree model, and wherein the decision tree model optimizes parameters of an iterative framework of the decision tree model and a weak learner regression tree by grid search during training.
5. The method according to any one of claims 2 to 4, wherein the size of the two-dimensional convolution kernels extracting the first and second spatially-correlated features is determined according to the resolution of the data, the higher the resolution is, the larger the size of the two-dimensional convolution kernels is, the shorter the time interval is, and the shorter the step size is, the one-dimensional convolution kernels extracting the first and second time-evolving features are determined according to the time interval of the data.
6. The method according to any one of claims 2 to 4, wherein the decision tree model calculates the contribution of each second spatial correlation feature and each second time evolution feature to the visibility prediction after each iteration is completed in the training process, and adjusts the weight of each feature according to the contribution.
7. A storage medium storing a computer program, characterized in that the computer program, when invoked by a processor for execution, implements the method of predicting airport visibility according to any one of claims 1 to 6.
8. An electronic device comprising a processor and a memory, said processor being communicatively connected to said memory via a data bus, said processor, when invoking a computer program in said memory, implementing a method for predicting airport visibility according to any one of claims 1-6.
9. A prediction system for airport visibility, comprising: the acquisition module is used for acquiring numerical weather forecast data and weather live data of the airport, which take effect in the airport and the nearby area; The data processing module is used for respectively extracting data related to visibility from the numerical weather forecast data and the weather live data to serve as numerical weather element forecast data and weather element live data; The differential feature extraction module is used for carrying out two-dimensional extraction on the numerical weather element forecast data and grid data in the weather element live data to obtain first space correlation features, and carrying out one-dimensional extraction on the numerical weather element forecast data and single-point time sequence weather data in the weather element live data to obtain first time evolution features; the visibility prediction module is used for inputting the first spatial correlation feature and the first time evolution feature into a pre-trained visibility prediction model to obtain a short-term visibility prediction result and an adjacent prediction result.
10. The machine field visibility prediction system according to claim 9, wherein the differential feature extraction module includes a convolution kernel matching module and an extraction module, the convolution kernel matching module is configured to determine a size of a two-dimensional convolution kernel for extracting the first spatial correlation feature according to a resolution of data, the size of the two-dimensional convolution kernel is larger as the resolution is higher, and determine a step size of a one-dimensional convolution kernel for extracting the first temporal evolution feature according to a time interval of the data, the step size of the one-dimensional convolution kernel is shorter as the time interval is shorter, and the extraction module is configured to extract the first spatial correlation feature through the corresponding two-dimensional convolution kernel and extract the first temporal evolution feature through the corresponding one-dimensional convolution kernel.

Description

Airport visibility prediction method, storage medium and related device Technical Field The invention relates to the technical field of airport meteorological element prediction, in particular to a prediction method and storage medium for airport visibility. Background The low visibility weather is a weather phenomenon occurring on the near ground layer, and serious vision impairment threatens the safety and efficiency of flight operation, so the requirement for establishing a quantitative and objective machine field visibility prediction model as a technical support is increasingly outstanding. Researchers at home and abroad also conduct a great deal of research and propose different solutions, mainly comprising a statistical forecasting method and a numerical forecasting method. The statistical forecasting method uses the relation between independent variables and dependent variables to establish a statistical model, mainly comprises linear regression, logistic regression and the like, has certain forecasting capability, but has limitation because of the non-linear characteristics of the generation and elimination of low-visibility weather, can simulate the microphysics and the thermodynamic processes of the generation and elimination of the low-visibility weather by utilizing a numerical mode, and gradually develops but is still not ideal. In actual business work, the machine field visibility prediction is mostly based on the change trend of weather situation and meteorological elements of live and numerical prediction, and the qualitative prediction of the visibility is made by combining historical experience. The existing prediction model based on the decision tree does not fully consider the structural difference of meteorological data, and a unified feature extraction mode is adopted, so that the feature extraction accuracy is insufficient, and the model convergence speed is low. Disclosure of Invention In order to overcome the defects of the prior art, the invention aims to provide the prediction method for the airport visibility, which can improve the accuracy of feature extraction and the convergence speed of a model, thereby improving the prediction efficiency and accuracy of the airport visibility. In order to solve the problems, the technical scheme adopted by the invention is as follows, namely, a prediction method of airport visibility, which comprises the following steps: Acquiring numerical weather forecast data and weather live data of an airport, which take effect in the airport and a nearby area; the method comprises the steps of respectively extracting data related to visibility from numerical weather forecast data and weather live data as numerical weather element forecast data and weather element live data; Two-dimensional extraction is carried out on grid type data in the digital weather element forecast data and the weather element live data to obtain first space correlation characteristics, and one-dimensional extraction is adopted on single-point time sequence weather data in the digital weather element forecast data and the weather element live data to obtain first time evolution characteristics; And inputting the first spatial correlation characteristic and the first time evolution characteristic into a pre-trained visibility prediction model to obtain a short-term visibility prediction result and an adjacent prediction result. Compared with the prior art, the prediction method for the visibility of the airport has the advantages that the first space correlation characteristic is obtained by two-dimensional extraction of grid type data in the acquired data, and the first time evolution characteristic is obtained by one-dimensional extraction of time sequence meteorological data in the data, so that the differential characteristic extraction of the meteorological data is realized, the extraction accuracy of the characteristics of the data with different structures is improved, the invalid search process in the model reasoning process is reduced, the convergence speed of the model is obviously improved, more accurate visibility prediction results can be output at a higher speed, and more efficient and reliable support is provided for airport management decisions. According to the airport visibility prediction method, the spatial correlation features and the time evolution features are input into the pre-trained visibility prediction model, and the short-term visibility prediction result and the close prediction result are obtained, wherein the visibility prediction model is trained through the following steps: Collecting global weather analysis data and airport historical weather live data for the last several years; The method comprises the steps of extracting data related to visibility of airports and nearby areas from global weather analysis data as weather element analysis data, and extracting data related to visibility from airport historical weather live data as airport weat