An intelligent visibility prediction method, device, equipment, medium and product

Abstract

The application discloses an intelligent visibility prediction method and device, equipment, medium and product, and relates to the field of intelligent visibility prediction. The method comprises the following steps: acquiring meteorological data and real visibility of a current time period; performing smooth processing on the real visibility to obtain current weighted visibility, and then performing difference calculation to obtain a weighted visibility difference sequence; inputting each factor data in the current meteorological data into a corresponding trained single-meteorological-factor prediction model to obtain meteorological data of a prediction time period; and inputting the predicted meteorological data and the current weighted visibility difference sequence into a trained weighted-difference visibility prediction model to obtain predicted weighted visibility of the prediction time period. The application fully considers the correlation of multiple factors by predicting meteorological data, combining weighted visibility processing and a difference sequence, and using a weighted-difference visibility prediction model, and also comprehensively considers time correlation and data smoothness, thereby improving the sensitivity to a low-visibility process and more accurately predicting visibility.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of visibility intelligent prediction, in particular to a visibility intelligent prediction method, device, equipment, medium and product. BACKGROUND

[0002] Meteorological conditions have a profound impact on the operation and development of many industries, among which visibility conditions are closely related to aviation safety, transportation efficiency, outdoor activity arrangement, etc. In the aviation field, visibility is directly related to the safety of aircraft take-off and landing, and low visibility may cause flight delays, cancellations, and even flight accidents; in the transportation sector, poor visibility can affect the driver's vision and increase the risk of traffic accidents; for outdoor activities such as sports events, tourism, etc., visibility will also affect the normal development and experience effect of the activities.

[0003] With the rapid progress of science and technology, visibility intelligent prediction technology, as an important branch of meteorological prediction, is gradually becoming a key technical means to ensure the safe and efficient operation of various industries. This technology aims to achieve accurate prediction of visibility and provide scientific basis for related decision-making. However, the visibility intelligent prediction model in related technology still has deficiencies. For example, in the case of rapid changes in visibility, the prediction accuracy of the model will be affected; for low-visibility conditions caused by some rare meteorological events, the model may lack sufficient data for learning and prediction. In particular in the aviation field, visibility below 4km will have some impact on aircraft take-off and landing, and the existing model still needs to be improved in terms of sensitivity in the low-visibility process, making it difficult to meet the strict requirements of the aviation industry for accurate prediction of low visibility.

[0004] Therefore, there is an urgent need for a more accurate and reliable visibility intelligent prediction method to improve the prediction ability of visibility under various meteorological conditions and ensure the safe operation of various industries. SUMMARY

[0005] The purpose of the present application is to provide a visibility intelligent prediction method, device, equipment, medium and product, which can effectively improve the accuracy and reliability of visibility prediction, especially enhance the sensitivity to low-visibility conditions.

[0006] To achieve the above-mentioned purpose, the present application provides the following solutions:

[0007] In a first aspect, the present application provides a visibility intelligent prediction method, comprising:

[0008] obtaining meteorological data and real visibility of a current time period; the meteorological data includes data of multiple meteorological factors; the current time period includes a current time and a first preset number of time points before the current time;

[0009] The real visibility of the current time period is smoothed by a weight visibility function to obtain the weight visibility of the current time period;

[0010] The weight visibilities of adjacent time points in the current time period are differentially calculated to obtain a weight visibility difference sequence of the current time period;

[0011] The data of each meteorological factor in the meteorological data of the current time period is respectively input into a corresponding trained single meteorological factor prediction model to obtain meteorological data of a prediction time period; the single meteorological factor prediction model is obtained by iteratively training a preset first machine learning model using the data of each meteorological factor in the meteorological data of a historical time period; the prediction time period includes a second preset number of time points after the current time point;

[0012] The meteorological data of the prediction time period and the weight visibility difference sequence of the current time period are input into a trained weight-difference visibility prediction model to obtain predicted visibility of the prediction time period; the weight-difference visibility prediction model is obtained by iteratively training a preset second machine learning model based on a sample training set; the sample training set includes the meteorological data of the historical time period, the real visibility, and the weight visibility difference sequence.

[0013] In a second aspect, the present application provides an intelligent visibility prediction device, comprising:

[0014] A data acquisition module is configured to acquire meteorological data and real visibility of a current time period; the meteorological data includes data of multiple meteorological factors; the current time period includes a current time point and a first preset number of time points before the current time point;

[0015] A weight visibility calculation module is configured to smooth the real visibility of the current time period by a weight visibility function to obtain the weight visibility of the current time period;

[0016] A weight visibility difference calculation module is configured to differentially calculate the weight visibilities of adjacent time points in the current time period to obtain a weight visibility difference sequence of the current time period;

[0017] A single meteorological factor prediction module is configured to input data of each meteorological factor in the meteorological data of the current time period into a corresponding trained single meteorological factor prediction model to obtain meteorological data of a prediction time period; the single meteorological factor prediction model is obtained by iteratively training a preset first machine learning model using the data of each meteorological factor in the meteorological data of a historical time period; the prediction time period includes a second preset number of time points after the current time point;

[0018] The visibility prediction module is used to input meteorological data for the prediction period and the weighted visibility difference sequence for the current period into a pre-trained weighted-difference visibility prediction model to obtain the predicted visibility for the prediction period. The weighted-difference visibility prediction model is obtained by iteratively training a preset second machine learning model based on a sample training set. The sample training set includes meteorological data for historical periods, actual visibility, and weighted visibility difference sequences.

[0019] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the visibility intelligent prediction method described in any one of the above.

[0020] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the visibility intelligent prediction method described in any one of the above descriptions.

[0021] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the visibility intelligent prediction method described above.

[0022] According to the specific embodiments provided in this application, this application has the following technical effects:

[0023] This application provides a method, device, equipment, medium, and product for intelligent visibility prediction. By acquiring meteorological data and actual visibility containing multiple meteorological factors for the current time period, it provides fundamental data support for subsequent predictions, solving the data source problem and achieving effective collection of raw information. Through the stationary processing of actual visibility using a weighted visibility function and differential calculation to obtain a weighted visibility difference sequence, it solves the problem of insufficient capture of dynamic changes during low visibility processes, achieving more accurate and timely visibility prediction. By inputting meteorological factor data into a pre-trained single meteorological factor prediction model to obtain predicted meteorological data, it solves the problem of inaccurate prediction of single meteorological factors, achieving accurate prediction of future data for each meteorological factor. By inputting the predicted meteorological data and the weighted visibility difference sequence into a weighted-difference visibility prediction model to obtain predicted visibility, it integrates meteorological factor prediction information and dynamic visibility change information, solving the problem of accurate visibility prediction based on multiple factors, and achieving effective prediction of future visibility. Attached Figure Description

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 An application environment diagram of an intelligent visibility prediction method in an embodiment of the present application;

[0026] Figure 2 A flowchart of an intelligent visibility prediction method provided in an embodiment of the present application;

[0027] Figure 3 A functional module diagram of an intelligent visibility prediction device provided in an embodiment of the present application;

[0028] Figure 4 A structural diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0029] Firstly, some technical terms involved in the embodiments of the present application are introduced.

[0030] There are two mainstream models of the visibility prediction method.

[0031] One is a fitting prediction model based on a mathematical method. The meteorological data is analyzed by a meteorological method, the meteorological law is converted into a mathematical algorithm, and fitting and prediction are performed by a mathematical method. In the related art, the fitting prediction model is used to study the airport visibility change trend under low visibility conditions. In the related art, a multiple regression model is established by analyzing the airport visibility data and taking the visibility as a dependent variable, and the airport visibility is predicted. However, this method needs to have certain physical priori knowledge, and the algorithm is calculated by converting the physical law into a specific mathematical algorithm. Since the actual environmental conditions are complex and changeable, the algorithm adaptability is poor, and the model needs to be constantly adjusted to ensure the prediction accuracy.

[0032] Second, the visibility intelligent prediction model. The visibility data and related meteorological data are taken as training samples, introduced into a machine learning model for training and obtaining a visibility prediction result. Related researchers propose an offshore visibility intelligent prediction method based on a temporal convolutional network (TCN) and transfer learning to solve the problem of less observation data and complex weather in offshore visibility prediction, which has achieved good results in the visibility prediction of the Qiongzhou Strait. Related researchers select wind speed, temperature, radiation, wind direction, air pressure, humidity, PM2.5 and PM10, a total of 8 data types, carry out atmospheric visibility prediction based on a deep belief network, train the predicted values of each meteorological parameter combined with visibility, extract the characteristics of the change of meteorological parameters and time change through the deep learning network, and finally obtain the future trend information of the visibility. This method only uses data as input, which will greatly simplify the model and reduce the computational complexity.

[0033] The visibility below 4km will have a certain impact on the take-off and landing of the aircraft. In order to improve the sensitivity of the machine learning model to the low visibility process, the present application improves and optimizes the visibility intelligent prediction model according to the actual demand, constructs a preset weight visibility function, and combines the time series method, the multiple regression method and the difference method to improve and optimize the visibility intelligent prediction model, so as to improve the sensitivity of the model to the low visibility process and improve the accuracy and reliability of the visibility prediction.

[0034] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0035] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail in combination with the drawings and specific embodiments.

[0036] The visibility intelligent prediction method provided by the embodiments of the present application can be applied to, for example Figure 1The application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data required by the server 104 to process. The data storage system can be set up separately, or integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the meteorological data and the real visibility of the current time period to the server 104. After the server 104 receives the meteorological data and the real visibility of the current time period, the server 104 smoothes the real visibility of the current time period by the weighted visibility function to obtain the weighted visibility of the current time period; difference calculation is performed on the weighted visibility of the adjacent time in the current time period to obtain the weighted visibility difference sequence of the current time period; the data of each meteorological factor in the meteorological data of the current time period is input into the corresponding trained single meteorological factor prediction model to obtain the meteorological data of the prediction time period; the single meteorological factor prediction model is obtained by iteratively training the preset first machine learning model using the data of each meteorological factor in the historical time period meteorological data; the prediction time period includes the second preset number of time points after the current time; the meteorological data of the prediction time period and the weighted visibility difference sequence of the current time period are input into the trained weighted-difference visibility prediction model to obtain the predicted visibility of the prediction time period; the weighted-difference visibility prediction model is obtained by iteratively training the preset second machine learning model based on the sample training set; the sample training set includes historical time period meteorological data, real visibility and weighted visibility difference sequence. The server 104 can feed back the predicted visibility of the prediction time period to the terminal 102. In addition, in some embodiments, the visibility intelligent prediction method can also be implemented by the server 104 or the terminal 102 alone, such as the terminal 102 can directly predict the visibility of the prediction time period for the meteorological data and the real visibility of the current time period, or the server 104 can obtain the meteorological data and the real visibility of the current time period from the data storage system and predict the visibility of the prediction time period for the meteorological data and the real visibility of the current time period.

[0037] Among them, the terminal 102 can be but not limited to various desktop computers, notebook computers, smart phones, tablet computers, Internet of Things devices and portable wearable devices. The server 104 can be implemented by an independent server or a server cluster composed of multiple servers, and can also be a cloud server.

[0038] In an exemplary embodiment, as shown in Figure 2 An intelligent visibility prediction method is provided, which is executed by a computer device, specifically by a terminal or a server, or by both, in the embodiments of the present application, the method is applied to the Figure 1The server 104 in the system 100 is taken as an example for illustration, including the following steps 201 to 205. Among them:

[0039] In step 201, meteorological data and real visibility in a current time period are obtained; the meteorological data includes data of multiple meteorological factors; the current time period includes a current time and a first preset number of time points before the current time; the meteorological factors include a temperature meteorological factor, a relative humidity meteorological factor, a wind speed meteorological factor, a wind direction meteorological factor and a barometric pressure meteorological factor.

[0040] In step 202, the real visibility in the current time period is smoothed by a weighted visibility function to obtain a weighted visibility in the current time period.

[0041] In step 203, the weighted visibilities of adjacent time points in the current time period are calculated by difference to obtain a weighted visibility difference sequence in the current time period.

[0042] In step 204, the data of each meteorological factor in the meteorological data in the current time period is respectively input into a corresponding trained single-meteorological-factor prediction model to obtain meteorological data in a prediction time period; the single-meteorological-factor prediction model is obtained by iteratively training a preset first machine learning model using the data of each meteorological factor in the meteorological data in a historical time period; the prediction time period includes a second preset number of time points after the current time.

[0043] In step 205, the meteorological data in the prediction time period and the weighted visibility difference sequence in the current time period are input into a trained weighted-difference visibility prediction model to obtain predicted visibility in the prediction time period; the weighted-difference visibility prediction model is obtained by iteratively training a preset second machine learning model based on a sample training set; the sample training set includes the meteorological data, the real visibility and the weighted visibility difference sequence in the historical time period.

[0044] By implementing the above steps 201 to 205, the application can effectively capture the changes of meteorological data and the dynamic trend of visibility, improve the sensitivity to low-visibility processes, significantly improve the accuracy and reliability of visibility prediction, and provide more scientific and accurate visibility prediction support for aviation, transportation, outdoor activities and other industries, effectively guarantee the safe operation and efficient decision-making of various industries.

[0045] In another exemplary embodiment of the application, step 205 specifically includes:

[0046] According to the meteorological data in the prediction time period and the weighted visibility difference sequence in the current time period, predicted weighted visibility difference data in the prediction time period are calculated.

[0047] The predicted weight visibility of the prediction time period is calculated according to the weight visibility of the current time period and the predicted weight visibility difference data of the prediction time period.

[0048] The predicted visibility of the prediction time period is obtained by the inverse weight visibility function based on the predicted weight visibility of the prediction time period; the inverse weight function is constructed by inversely deducing the preset weight visibility function.

[0049] In another exemplary embodiment of the present application, the expression of the weight visibility function is:

[0050]

[0051] wherein W represents the weight visibility; V represents the real visibility; a, b, c, d, e and f respectively represent the first weight coefficient, the second weight coefficient, the third weight coefficient, the fourth weight coefficient, the fifth weight coefficient and the sixth weight coefficient; a = 1, b = 0, c = 0.25, d = 4.5, e = 1 / 14, f = 59 / 7.

[0052] In another exemplary embodiment of the present application, the expression of the predicted weight visibility difference data of the prediction time period is calculated according to the meteorological data of the prediction time period and the weight visibility difference sequence of the current time period.

[0053]

[0054] wherein DW t+1 represents the predicted weight visibility difference data of the t+1 moment; f() represents the mapping function; T t+1 represents the temperature meteorological factor of the t+1 moment; RH t+1 represents the relative humidity meteorological factor of the t+1 moment; S t+1 represents the wind speed meteorological factor of the t+1 moment; D t+1 represents the wind direction meteorological factor of the t+1 moment; P t+1 represents the air pressure meteorological factor of the t+1 moment; DW t and DW t-j respectively represent the weight visibility difference data of the t moment and the weight visibility difference data of the t-j moment, and j represents the first preset number.

[0055] In another exemplary embodiment of the present application, the preset first machine learning model and the second machine learning model are both random forest algorithm models.

[0056] In another exemplary embodiment of the present application, an intelligent visibility prediction method is provided, which specifically comprises the following steps:

[0057] S1: data preparation and processing.

[0058] S1-1: Study data, taking weather station observation data as an example, filter the weather observation data of a certain station from January 1, 2016 to December 31, 2023, mainly including visibility V, temperature T, relative humidity RH, wind speed S, wind direction D, air pressure P, etc.

[0059] S1-2: The collected data will have different degrees of missing, and the sample set with missing data can be deleted after the sample set is constructed.

[0060] S1-3: Analyze the meteorological elements that have a greater impact on visibility as input for the prediction model.

[0061] S2: Construct a preset weight visibility function.

[0062] S2-1: According to the visibility requirements, the 0-50km visibility can be divided into three sections, 0-6km, 6-22km, and 22-50km, and the visibility from 6km to 50km is reduced in the prediction weight. Visibility below 4km will have some impact on helicopter takeoff and landing, considering the 2km error, set the first interval to 0-6km; visibility is mostly within 22km, set the second and third intervals.

[0063] S2-2: Use a large amount of data to train the preset weight visibility function:

[0064]

[0065] Where W represents the weight visibility; V represents the true visibility; a, b, c, d, e, and f represent the first, second, third, fourth, fifth, and sixth weight coefficients, respectively.

[0066] S2-3: The weight coefficients are continuously adjusted during the training process, when the weight coefficients a and b are 1 and 0, respectively, and c and d are 0.25 and 4.5, respectively, and e and f are 1 / 14 and 59 / 7, respectively, the error between the predicted visibility and the true visibility is smaller.

[0067] S2-4: According to different requirements, set the visibility interval, the preset weight visibility function, and the weight coefficient, repeat steps S2-1, S2-2, and S2-3.

[0068] S3: Construct a single meteorological factor prediction model.

[0069] S3-1: Convert time series data into training samples with input and output characteristics, mainly input 24 hours of temperature T (or relative humidity RH or wind speed S or wind direction D or air pressure P) and output the 25th hour of temperature T (or relative humidity RH or wind speed S or wind direction D or air pressure P), and so on.

[0070] S3-2: Train the training sample set based on the random forest algorithm. The random forest is composed of multiple decision trees, and parameters such as the number of trees and the depth of the trees are set. Each decision tree learns the corresponding sub-sample and feature subset, and generates a prediction model.

[0071] S3-3: Input the data of the previous 1 to 24 hours into the training model, and output the temperature T t+1 (or relative humidity RH t+1 , or wind speed S t+1 , or wind direction D t+1 , or air pressure P t+1 ) at the t+1 time.

[0072] S4: Construct a weighted visibility difference data prediction model.

[0073] S4-1: Construct a training sample set. The input of the training sample set is the meteorological parameters of the 25th time period and the difference of the weighted visibility of the 1st to 24th time periods, and the output is the difference DW t+1 of the weighted visibility of the 25th and 24th time periods, and so on. The difference operation is as follows:

[0074] DW t+1 = W t+1 -W t .

[0075] Wherein, DW t+1 represents the weighted visibility difference data at the t+1 time; W t+1 , W t represent the weighted visibility at the t+1 time and the t time, respectively.

[0076] S4-2: Train the training sample set based on the random forest algorithm, and each decision tree carries the weight parameter, records the importance and mutual relationship of the input features, and the weight is also updated in the model training process, and the optimal prediction model is generated.

[0077] S4-3: Input the meteorological parameters (T t+1 , RH t+1 , W t+1 , D t+1 , P t+1 ) predicted by the single meteorological factor prediction model at the t+1 time and the difference (DW t , …, DW t-23 ) of the weighted visibility from the t time to the t-23 time of the current time period into the training model for prediction, and output the difference DW t+1 of the weighted visibility at the t+1 time. Then the weighted visibility at the t+1 time is:

[0078] W t+1 = DW t+1 +W t .

[0079] wherein, W t+1 represents the weight visibility at the t+1 time; W t represents the weight visibility at the t time.

[0080] Based on the inverse weight visibility function as shown below. The visibility at the t+1 time can be obtained:

[0081]

[0082] S5: Constructing 24-hour visibility intelligent prediction model.

[0083] S5-1: Constructing 24-hour visibility intelligent prediction model based on weight visibility difference data prediction model with 1 hour as a unit.

[0084] S5-2: Inputting the meteorological parameters (T t+1 , RH t+1 , W t+1 , D t+1 , P t+1 ) predicted by the single meteorological factor prediction model at the t+1 time and the difference (DW t , …, DW t-23 ) of the weight visibility at the t time to the t-23 time of the current time period into the training model to predict, output the difference DW t+1 of the weight visibility at the t+1 time, and further obtain the visibility V t+1 at the t+1 time. In this way, inputting the meteorological parameters (T t+24 , RH t+24 , W t+24 , D t+24 , P t+24 ) predicted by the single meteorological factor prediction model at the t+24 time period and the difference (DW t , …, DW t-23 ) of the weight visibility at the t time to the t-23 time into the training model to predict, output the difference DW t+24 of the weight visibility at the t+24 time, and further obtain the visibility V t+24 at the t+24 time. Through the above operation, the 24-hour visibility, i.e., V t+1 , V t+2 , …, V t+23 , V t+24 , can be predicted.

[0085] S6: Visibility intelligent prediction model evaluation.

[0086] S6-1: The visibility intelligent prediction model is tested and evaluated by using the root mean square error (RMSE). The model prediction results are compared with the field observation results. The root mean square error is:

[0087]

[0088] wherein N represents the total number of meteorological data samples; V n represents the predicted visibility of the nth meteorological data sample; and V on represents the true visibility of the nth meteorological data sample.

[0089] S6-2: The model is evaluated.

[0090] S6-2-1: The training sample mainly inputs the meteorological parameters (T, RH, S, D, P) of a certain time period and outputs the visibility V of a certain time period. Based on the random forest algorithm, a multivariate visibility intelligent prediction model is constructed, and the training sample set is trained. The meteorological parameters (T t+1 , RH t+1 , S t+1 , D t+1 , P t+1 ) of the t+1 time predicted by the single meteorological factor prediction model are input into the training model, and the visibility V t+1 of the t+1 time is output.

[0091] S6-2-2: The training sample mainly inputs the visibility of 1-24 hours and outputs the visibility V of 25 hours, and so on. Based on the random forest algorithm, a time series visibility intelligent prediction model is constructed, and the training sample set is trained. The visibility (V t , …, V t-23 ) from the t time to the t-23 time is input into the training model, and the visibility V t+1 of the t+1 time is output.

[0092] S6-2-3: The multivariate time series visibility intelligent prediction model is constructed by combining S5-2-1 and S5-2-1, and the training sample set is trained. The meteorological parameters (T t+1 , RH t+1 , S t+1 , D t+1 , P t+1 ) of the t+1 time predicted by the single meteorological factor prediction model and the visibility (V t , …, V t-23 ) from the t time to the t-23 time are input into the training model, and the visibility V t+1 of the t+1 time is output.

[0093] S6-2-4: Construct a weight-based multivariate time sequence visibility intelligent prediction model, and train the training sample set. The meteorological parameters (T t+1 , RH t+1 , S t+1 , D t+1 , P t+1 ) predicted by the single meteorological factor prediction model at the t+1 time and the weight visibility (W t , …, W t-23 ) from the t time to the t-23 time are input into the training model, and the weight visibility W t+1 at the t+1 time is output, and the visibility V t+1 is obtained through the inverse weight model.

[0094] S6-2-5: Construct a difference-based multivariate time sequence visibility intelligent prediction model, and train the training sample set. The meteorological parameters (T t+1 , RH t+1 , S t+1 , D t+1 , P t+1 ) predicted by the single meteorological factor prediction model at the t+1 time and the difference (D t , …, D t-23 ) of the visibility from the t time to the t-23 time are input into the training model, and the visibility difference D′ t+1 at the t+1 time is output, and the visibility V t+1 is obtained.

[0095] D′ t+1 = V t+1 -V t .

[0096] S6-2-6: Weight-difference visibility prediction model.

[0097] S6-3: Test and evaluation.

[0098] The evaluation method is used to compare and analyze the results of the six evaluation models (multivariate visibility intelligent prediction model, time sequence visibility intelligent prediction model, multivariate time sequence visibility intelligent prediction model, weight-based multivariate time sequence visibility intelligent prediction model, difference-based multivariate time sequence visibility intelligent prediction model, and weight-difference visibility prediction model). The weight-difference visibility prediction model of the present application has the smallest error.

[0099] The application further provides an application scenario of the visibility intelligent prediction method. Specifically, the visibility intelligent prediction method provided in the embodiment can be applied in an air traffic management scenario. The air traffic management scenario includes a meteorological data acquisition link, a visibility prediction link, and a flight scheduling decision link. Meteorological data enters the visibility prediction link from the meteorological data acquisition link, is processed by the visibility intelligent prediction method provided in the application, and obtains predicted visibility data of a prediction time period, and enters the flight scheduling decision link. The visibility intelligent prediction method provided in the embodiment belongs to the visibility prediction link in air traffic management. Specifically, through comprehensive analysis and accurate prediction of multiple meteorological factors, more reliable visibility basis is provided for flight scheduling, and aviation traffic safety and efficient operation are facilitated.

[0100] Based on the same inventive concept, the embodiment of the application further provides a visibility intelligent prediction device for implementing the visibility intelligent prediction method described above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more visibility intelligent prediction device embodiments provided below can refer to the limitations of the visibility intelligent prediction method described above, and will not be repeated here.

[0101] In one exemplary embodiment, as shown in Figure 3 a visibility intelligent prediction device is provided, comprising:

[0102] The data acquisition module 301 is configured to acquire meteorological data and real visibility of a current time period. The meteorological data includes data of multiple meteorological factors. The current time period includes a current time and a first preset number of time points before the current time.

[0103] The weight visibility calculation module 302 is configured to perform smooth processing on the real visibility of the current time period by using a weight visibility function, to obtain weight visibility of the current time period.

[0104] The weight visibility difference calculation module 303 is configured to perform difference calculation on the weight visibility of adjacent time points in the current time period, to obtain a weight visibility difference sequence of the current time period.

[0105] The single meteorological factor prediction module 304 is configured to input the data of each meteorological factor in the meteorological data of the current time period into a corresponding trained single meteorological factor prediction model, to obtain meteorological data of a prediction time period. The single meteorological factor prediction model is obtained by iteratively training a preset first machine learning model using the data of each meteorological factor in the meteorological data of a historical time period. The prediction time period includes a second preset number of time points after the current time.

[0106] The prediction visibility calculation module 305 is configured to input the meteorological data of the prediction time period and the weighted visibility difference sequence of the current time period into the trained weighted-difference visibility prediction model to obtain the predicted visibility of the prediction time period; the weighted-difference visibility prediction model is obtained by iteratively training a preset second machine learning model based on a sample training set; the sample training set includes the meteorological data, the real visibility and the weighted visibility difference sequence of the historical time period.

[0107] In an exemplary embodiment, a computer device is provided, which can be a server or a terminal, and an internal structure diagram of the computer device can be as shown in Figure 4 The computer device includes a processor, a memory, an input / output interface (I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The database of the computer device is configured to store meteorological data and visibility processing data. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with external terminals through network connection. The computer program is executed by the processor to implement an intelligent visibility prediction method.

[0108] Those skilled in the art can understand that Figure 4 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, which includes a memory and a processor, and the memory stores a computer program. The processor executes the computer program to implement the steps in the above method embodiments.

[0109] In an exemplary embodiment, a computer readable storage medium is provided, which stores a computer program. The computer program is executed by a processor to implement the steps in the above method embodiments.

[0110] In an exemplary embodiment, a computer program product is provided, which includes a computer program. The computer program is executed by a processor to implement the steps in the above method embodiments.

[0111] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0112] It can be understood by those skilled in the art that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing related hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments of each method. Any reference to memory, database or other medium used in each embodiment provided by the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ReadOnly Memory, ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive random access memory (Resistive Random Access Memory, ReRAM), magnetoresistive random access memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric memory (Ferroelectric Random Access Memory, FRAM), phase change memory (Phase Change Memory, PCM), graphene memory, etc. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), etc.

[0113] The database involved in each embodiment provided by the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in each embodiment provided by the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0114] Any technical features in the above embodiments can be combined, and for the sake of brevity, not all possible combinations are described above, however, it should be understood that the application encompasses all possible combinations of the technical features unless such a combination is not technically possible.

[0115] The principles and implementation manners of the present application are described herein by using specific examples, and the above embodiments are only used to help understand the method of the present application and its core idea; meanwhile, according to the idea of the present application, a person skilled in the art can make changes in specific implementation manners and application scopes. In conclusion, the content of the present specification should not be understood as a limitation of the present application.

Claims

1. A method for intelligent visibility prediction, characterized in that, The intelligent visibility prediction method includes: Acquire meteorological data and actual visibility for the current time period; the meteorological data includes data on multiple meteorological factors; the current time period includes the current moment and the first preset number of moments before the current moment; The weighted visibility for the current time period is obtained by smoothing out the true visibility using a weighted visibility function; the expression for the weighted visibility function is as follows: ； in, Indicates the visibility of weights; Indicates true visibility; , , , , and These represent the first weighting coefficient, the second weighting coefficient, the third weighting coefficient, the fourth weighting coefficient, the fifth weighting coefficient, and the sixth weighting coefficient, respectively; a=1, b=0, c=0.25, d=4.5, e=1 / 14, f=59 / 7; The weighted visibility difference sequence for the current time period is obtained by performing differential calculation on the weighted visibility of adjacent times within the current time period. The data of each meteorological factor in the meteorological data of the current time period is input into the corresponding pre-trained single meteorological factor prediction model to obtain the meteorological data for the prediction time period; the single meteorological factor prediction model is obtained by iteratively training a preset first machine learning model using the data of each meteorological factor in the meteorological data of historical time periods; the prediction time period includes the second preset number of times after the current time. Meteorological data for the predicted time period and the weighted visibility difference sequence for the current time period are input into a pre-trained weighted-difference visibility prediction model to obtain the predicted visibility for the predicted time period. The weighted-difference visibility prediction model is obtained by iteratively training a preset second machine learning model based on a sample training set. The sample training set includes meteorological data for historical time periods, actual visibility, and weighted visibility difference sequences.

2. The visibility intelligent prediction method according to claim 1, characterized in that, The meteorological data for the forecast period and the weighted visibility difference sequence for the current period are input into a pre-trained weighted-difference visibility prediction model to obtain the predicted visibility for the forecast period, specifically including: Based on the meteorological data for the forecast period and the weighted visibility difference sequence for the current period, calculate the forecast weighted visibility difference data for the forecast period. Calculate the predicted weight visibility for the predicted time period based on the difference between the weight visibility of the current time period and the predicted weight visibility of the predicted time period. Based on the predicted weight visibility for the predicted time period, the predicted visibility for the predicted time period is obtained through the inverse weight visibility function; the inverse weight visibility function is constructed by reverse derivation of the preset weight visibility function.

3. The intelligent visibility prediction method according to claim 1, characterized in that, Based on the meteorological data for the forecast period and the weighted visibility difference sequence for the current period, the expression for calculating the forecast weighted visibility difference data for the forecast period is as follows: ； in, This represents the predicted weighted visibility difference data at time t+1; Represents a mapping function; This represents the temperature meteorological factor at time t+1. The relative humidity meteorological factor represents the relative humidity at time t+1. The wind speed meteorological factor represents the wind speed at time t+1. The wind direction meteorological factor represents the wind direction at time t+1. This represents the barometric meteorological factor at time t+1; and Let represent the weighted visibility difference data at time t and the weighted visibility difference data at time tj, respectively, where j represents the first preset number.

4. The intelligent visibility prediction method according to claim 1, characterized in that, Both the preset first and second machine learning models are random forest algorithm models.

5. The intelligent visibility prediction method according to claim 1, characterized in that, The meteorological factors include temperature, relative humidity, wind speed, wind direction, and air pressure.

6. A visibility intelligent prediction device, characterized in that, The visibility intelligent prediction device applies the visibility intelligent prediction method according to any one of claims 1-5, wherein the visibility intelligent prediction device comprises: The data acquisition module is used to acquire meteorological data and actual visibility for the current time period; the meteorological data includes data of multiple meteorological factors; the current time period includes the current moment and the first preset number of moments before the current moment; The weighted visibility calculation module is used to smooth the actual visibility of the current time period through the weighted visibility function to obtain the weighted visibility of the current time period. The weighted visibility difference calculation module is used to perform difference calculation on the weighted visibility of adjacent times in the current time period to obtain the weighted visibility difference sequence of the current time period. The single meteorological factor prediction module is used to input the data of each meteorological factor in the meteorological data of the current time period into the corresponding pre-trained single meteorological factor prediction model to obtain the meteorological data for the prediction time period. The single meteorological factor prediction model is obtained by iteratively training a preset first machine learning model using the data of each meteorological factor in the meteorological data of historical time periods. The prediction time period includes the second preset number of times after the current time. The visibility prediction module is used to input meteorological data for the prediction period and the weighted visibility difference sequence for the current period into a pre-trained weighted-difference visibility prediction model to obtain the predicted visibility for the prediction period. The weighted-difference visibility prediction model is obtained by iteratively training a preset second machine learning model based on a sample training set. The sample training set includes meteorological data for historical periods, actual visibility, and weighted visibility difference sequences.

7. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the visibility intelligent prediction method according to any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the visibility intelligent prediction method according to any one of claims 1-5.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the visibility intelligent prediction method according to any one of claims 1-5.

Images