Airport passenger flow multi-factor fusion prediction method based on causal inference

Abstract

The application relates to an airport passenger flow multi-factor fusion prediction method based on causal inference, belongs to the technical field of airport passenger flow prediction, and solves the problem that the change of passenger traffic volume brings difficulties to airport resource allocation in the prior art, and comprises the following steps: acquiring input data information; performing Granger causal inference test on airport passenger flow; obtaining a Granger causal inference result of multiple factors; dividing the time sequence of input features for predicting airport passenger flow; establishing an attention mechanism in a time period and obtaining local time sequence information; establishing an attention mechanism between time periods and obtaining global time sequence information; fusing the obtained local time sequence information and global time sequence information with the Granger causal inference result to obtain a prediction value of the airport passenger flow; calculating the error between the prediction value and an actual value; obtaining a trained airport passenger flow prediction model; and inputting measured real-time data into the trained airport passenger flow prediction model to obtain airport passenger flow prediction information.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of airport passenger flow prediction, in particular to an airport passenger flow multi-factor fusion prediction method based on causal inference. BACKGROUND

[0002] With the increase of travel demand, it brings the increase of civil aviation passenger traffic, which intensifies the imbalance between airport resources and airport passenger flow, and brings great pressure to the internal resource allocation of the airport and the surrounding traffic system. Forecasting the passenger flow in the airport in advance is one of the most effective methods to balance the airport resources and passenger demand. Through the prediction of passenger flow, the airport management can better dispatch traffic resources, allocate and arrange resources in advance to cope with the emergency and passenger flow pressure in peak hours. For passengers, it reduces the waiting time and improves the travel experience; for the airport, it reduces the operating cost and improves the resource utilization. Therefore, it is of practical significance to predict the airport passenger flow in advance.

[0003] If the predicted airport passenger flow value is too low, it will cause passengers to stay in the airport, cause public opinion and surrounding traffic paralysis and other problems. If the predicted airport passenger flow value is too high, it will cause waste of resources and economic loss, such as idle equipment resources and low investment return. Therefore, it is necessary to accurately predict the airport passenger flow. At the same time, the prediction of airport passenger flow itself also faces a series of challenges. The airport passenger flow is affected by multiple related factors, such as flight delay, weather conditions, sudden regulation, etc. The influencing factors are complex. And these influencing factors interact and correlate with each other, making the airport passenger flow data prone to sudden fluctuations in a short time. These reasons make the airport passenger flow prediction problem complex and difficult.

[0004] However, in the existing airport passenger flow prediction method, the multi-influencing factor coupling characteristics of passenger flow are ignored. In order to make the airport passenger flow prediction method perform better and better meet the actual needs, the technical field needs an improved prediction method that can fully grasp the multi-factor coupling strength of passenger flow data and thus improve the prediction accuracy. SUMMARY

[0005] In view of the above problems, the present application provides an airport passenger flow multi-factor fusion prediction method based on causal inference.

[0006] According to the embodiment of the present application, an airport passenger flow multi-factor fusion prediction method based on causal inference is provided, comprising the following steps:

[0007] The airport passenger flow multi-factor fusion prediction method based on causal inference, characterized in that it comprises the following steps:

[0008] Step S0, obtaining input data information, including obtaining passenger flow data, flight schedule, carbon dioxide concentration, PM2.5 concentration, temperature and humidity information through database and sensor, all information in the input data information is time series data;

[0009] Step S1, based on the obtained input data information, performing Granger causality inference test on the airport passenger flow, constructing multi-factor causal correlation, and obtaining input features for predicting the airport passenger flow;

[0010] Step S2, judging whether the flight schedule, carbon dioxide concentration, PM2.5 concentration, temperature and humidity information in the input data information obtained in step S0 have been subjected to Granger test and analysis, if yes, taking all the obtained input features as the Granger causality inference result of the multi-factor, and performing the next step, otherwise returning to step S1 to perform Granger causality inference test on the remaining influencing factors in the input data information;

[0011] Step S3, dividing the time series of the input features to obtain new time series;

[0012] Step S4, extracting features in each time period in the obtained new time series, establishing an attention mechanism in the time period and obtaining local time series information;

[0013] Step S5, extracting features between time periods in the obtained new time series, establishing an attention mechanism between time periods and obtaining global time series information to capture global correlation;

[0014] Step S6, fusing the obtained local time series information and global time series information with the Granger causality inference result of step S2 to realize the effect of weighting, and obtaining the predicted value of the airport passenger flow;

[0015] Step S7, calculating the error between the obtained predicted value and the true value of the passenger flow data obtained in step S0, and updating all weight parameters in the neural network using gradient descent method;

[0016] Step S8, comparing the obtained error with a predetermined threshold, and comparing the number of iterations with a maximum number of iterations, when one of the two conditions that the error is less than the predetermined threshold and the number of iterations reaches the maximum number of iterations is satisfied, the training model ends, and a trained airport passenger flow prediction model is obtained, otherwise returning to step S3 to re-divide the time series in different scales S;

[0017] Step S9, inputting the measured real-time data into the obtained trained airport passenger flow prediction model to obtain airport passenger flow prediction information, which is used for airport resource scheduling and allocation.

[0018] Optionally, the step S1 specifically comprises the following steps:

[0019] Step S1.1, selecting features from the flight schedule, carbon dioxide concentration, PM2.5 concentration, temperature and humidity information in the obtained input data information, constructing a multivariate regression model of airport passenger flow as:

[0020]

[0021] wherein Y t represents the airport passenger flow at time t, represents the value of the feature at time t-q, β p ,η t ,γ q represents the regression parameter and is obtained by training data, p represents the variable at time p, and q represents the variable at time q;

[0022] Step S1.2, constructing a univariate regression model of airport passenger flow by adding the past values of the influencing factors and performing joint regression analysis as:

[0023] Y t = α0+ α1Y t-1 +…+α p Y t-p +ε t

[0024] wherein α p ,ε t represents the regression parameter and is obtained by training data;

[0025] Step S1.3, performing F test on the established multivariate regression model and univariate regression model of airport passenger flow to determine whether the past values of the features X t can significantly improve the prediction ability of the airport passenger flow Y t at time t.

[0026] Optionally, the step S1.3 specifically comprises the following steps:

[0027] Step S1.3.1, performing F test on the established multivariate regression model and univariate regression model of airport passenger flow to obtain a test value:

[0028]

[0029] wherein RSS r is the residual sum of squares of the univariate autoregressive model, RSS ur is the residual sum of squares of the multivariate regression model, q is the number of introduced lag terms, n is the sample size, and k is the number of explanatory variables in the model;

[0030] Step S1.3.2, analyzing the result obtained in step S1.3.1, if the obtained test value F is greater than a preset threshold value, it indicates that the current feature X t Granger causes Y t , otherwise, it indicates that the feature has no effect on prediction, and it is deleted.

[0031] Optionally, step S3 specifically comprises:

[0032] The obtained time series data of input features X∈R H×d is taken as input, where H represents the length of the time series, d represents the dimension of the feature, and R H×d represents a set of time series data, and data division is performed using a scale of length S to obtain a new set of time series:

[0033] (X1,...,X i ...,X P )∈R P×S×d

[0034] where X i represents the input data of the i-th time period and X i ∈R S×d , P represents the number of time periods obtained after dividing the time series data, each time period as a new unit.

[0035] Optionally, step S4 specifically comprises the following steps:

[0036] Step S4.1, for the input data X i ∈R P of the i-th time period in the new time series (X1,...,X i ...,X S×d ) after division, encoding in the feature dimension d to obtain:

[0037]

[0038] where, represents the new input data in the i-th time period after encoding in the feature dimension d, d m represents encoding the feature dimension d to d m ;

[0039] Step S4.2, performing linear transformation on to obtain a variable in the attention mechanism:

[0040]

[0041] where, denotes the key vector in the time period, V i intra denotes the value vector in the time period;

[0042] Step S4.3, using trainable variables combining the before-and-after time series information in the time period, the input data X i in the i-th time period is calculated:

[0043]

[0044] Step S4.4, performing steps S4.1-S4.3 on the input data of each time period in the divided new time series to obtain the attention information as local time series information.

[0045] Optionally, step S5 specifically includes the following steps:

[0046] Step S5.1, encoding the divided new time series X P×S×d in the feature dimension d to d m ', and the mapped data is:

[0047]

[0048] where d m '= Sxd;

[0049] Step S5.2, performing self-attention operation between time periods to model the airport passenger flow data based on the correlation between time periods, and obtaining the variable between time periods through linear mapping of X inter

[0050]

[0051] where Q inter denotes the query vector between time periods, K inter denotes the key vector between time periods, V inter denotes the value vector between time periods;

[0052] Step S5.3, obtaining the attention information between time periods:

[0053]

[0054] Step S5.4, performing steps S5.1-5.3 between all time periods of the divided new time series to obtain the attention information between time periods as global time series information.

[0055] ​The application provides an airport passenger flow multi-factor fusion prediction method based on causal inference, comprising: (1) causal coupling strength inference: the airport passenger flow is affected by multiple factors, in the patent, the flight schedule information table, co2 concentration, pm2.5 concentration, temperature, humidity and historical airport passenger flow data are considered, all the data are time series data, wherein the flight schedule information table refers to the number of aircrafts arriving at the airport per unit time. The causal correlation of multiple factors is constructed by using Granger causal inference test, and the multi-factor coupling strength information is obtained. (2) Time sequence feature block extraction: the time sequence is divided into patch of subsequence level, the patch is used as the input of the Transform model, the information in the patch is extracted by using the self-attention mechanism, and the local time sequence information is obtained; then the self-attention mechanism is used for feature extraction between the patches, and the global time sequence information is obtained. (3) Information fusion, the obtained time sequence information and Granger causal information are fused, and a prediction result containing causal coupling strength inference information is output. In the case that the passenger flow is affected by many dynamic and static elements and the correlation between the elements is complex, the coupling strength of the airport passenger flow under multiple factors is inferred, and the accurate airport passenger flow prediction result is obtained.

[0056] Compared with the prior art, the airport passenger flow multi-factor fusion prediction method based on causal inference provided by the embodiment of the application has at least the following beneficial effects:

[0057] (1) The characteristics of the coupling of multiple influencing factors of the airport passenger flow are considered, the coupling strength of the multiple factors is accurately extracted through Granger causal analysis, and an accurate airport passenger flow prediction value is obtained.

[0058] (2) By means of time sequence division, the local fluctuation of the airport passenger flow is extracted by taking the time block as a new unit, and the global trend of the data is captured through feature extraction between the time blocks.

[0059] (3) In practical application, the passenger flow in the airport is effectively predicted, so that the balance between airport resources and passenger transport demand is realized.

[0060] (4) By providing an accurate airport passenger flow prediction value, it is beneficial to reasonably and efficiently allocate resources. BRIEF DESCRIPTION OF DRAWINGS

[0061] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the drawings needed in the embodiments will be briefly introduced as follows, and the features and advantages of the application can be more clearly understood by referring to the drawings. The drawings are schematic and should not be understood as any limitation on the application. For those skilled in the art, other drawings can be obtained without creative labor on the basis of these drawings.

[0062] Figure 1 A principle schematic diagram of an airport passenger flow multi-factor fusion prediction method based on causal inference provided for an embodiment of the present application.

[0063] Figure 2 A flowchart of an airport passenger flow multi-factor fusion prediction method based on causal inference provided for an embodiment of the present application. DETAILED DESCRIPTION

[0064] In order to more clearly understand the above-mentioned purposes, features and advantages of the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0065] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, therefore, the scope of protection of the present application is not limited by the specific embodiments disclosed below.

[0066] The airport passenger flow multi-factor fusion prediction method based on causal inference provided for an embodiment of the present application will be described in detail below with reference to the accompanying drawings.

[0067] As shown in Figure 1 and Figure 2 The airport passenger flow multi-factor fusion prediction method based on causal inference provided for an embodiment of the present application includes the following steps.

[0068] Step S0, obtaining input data information, including obtaining passenger flow data, flight schedule, carbon dioxide concentration, PM2.5 concentration, temperature, humidity and other information through a database and a sensor. All data in the input data information are time series data. Alternatively, the input data information can be historical data information, which is used for model construction and training.

[0069] Step S1, based on the obtained input data information, performing Granger causal inference test on the airport passenger flow to construct multi-factor causal correlation. The Granger causal inference test determines the causal relationship through a regression model, including establishing a multivariate regression model and a univariate regression model, and step S1 specifically includes the following steps.

[0070] Step S1.1, selecting features from the flight schedule, carbon dioxide concentration, PM2.5 concentration, temperature and humidity information in the obtained input data information, and constructing a multivariate regression model of the airport passenger flow as follows:

[0071]

[0072] wherein Y tYt represents the airport passenger flow at time t, t-p Yt-p represents the airport passenger flow at time t-p, βt-q represents the value of the feature at time t-q, p ,η t ,γ q represent regression parameters and are trained from data, p represents the variable at time p, and q represents the variable at time q.

[0073] Step S1.2, by adding the past values of the influencing factors, a joint regression analysis is performed to construct a univariate regression model of the airport passenger flow as follows:

[0074] Y t = α0+ α1Y t-1 +…+ α p Y t-p +ε t

[0075] wherein, α p ,ε t represent regression parameters and are trained from data.

[0076] Step S1.3, F test is performed on the established multivariate regression model and univariate regression model of the airport passenger flow to determine whether the past values of the feature X t can significantly improve the prediction ability of Y t . The step S1.3 specifically includes the following steps.

[0077] Step S1.3.1, F test is performed on the established multivariate regression model and univariate regression model of the airport passenger flow to obtain a test value as follows:

[0078]

[0079] wherein, RSS r is the residual sum of squares of the univariate autoregressive model, RSS ur is the residual sum of squares of the multivariate regression model, q is the number of introduced lag terms, n is the sample size, and k is the number of explanatory variables in the model.

[0080] Step S1.3.2, the result obtained in step S1.3.1 is analyzed, if the obtained test value F is greater than a preset critical value, it indicates that the current feature X t Granger causes Y t , the current feature is taken as an input feature for predicting the airport passenger flow, otherwise, it indicates that the current feature has no effect on the prediction and is deleted.

[0081] Step S2: Determine whether all features in the input data information obtained in step S0, such as flight schedules, carbon dioxide concentration, PM2.5 concentration, temperature and humidity information, have been subjected to Granger tests and analyses. If so, use all the input features of the predicted airport passenger flow as the results of multi-factor Granger causal inference and proceed to the next step. Otherwise, return to step S1 to perform Granger causal inference tests on the remaining influencing factors in the input data information.

[0082] Step S3: Divide the time series to obtain a new time series. Use the obtained time series data X∈R of the input features for predicting airport passenger traffic. H×d As input, where H represents the length of the time series, d represents the dimension of the feature, and R... H ×d The set of time series data is divided using a scale of length S to obtain a new set of time series data.

[0083] (X1,...,X i ...,X P )∈R P×S×d

[0084] Among them, X i Let X represent the input data for the i-th time period and X i ∈R S×d P represents the number of time periods obtained after dividing the time series data. Each time period is treated as a new unit, called a Patch. Self-attention within each time period needs to be established within each Patch to obtain local temporal information.

[0085] Step S4: Extract features from each time segment of the newly divided time series, establish an attention mechanism within each time segment, and obtain local temporal information. Step S4 specifically includes the following steps.

[0086] Step S4.1: For the newly partitioned time series (X1,...,X...), i ...,X P The input data X for the i-th time period in ) i First, we encode the feature dimension d to obtain:

[0087]

[0088] in, This represents the new input data within the i-th time period after encoding on feature dimension d, where d m This indicates that the feature dimension d is encoded into d. m .

[0089] Step S4.2, then linear transformation is performed on to obtain the attention mechanism in

[0090]

[0091] wherein, represents the key vector in the time period, V i intra represents the value vector in the time period.

[0092] Step S4.3, using trainable variables to combine the front and rear time series information in the time period, then the attention information in the i-th time period X i is calculated:

[0093]

[0094] Step S4.4, performing steps S4.1-4.3 on the input data of each time period in the divided new time series to obtain the attention information as the local time series information.

[0095] Step S5, feature extraction is performed between the time periods in the divided new time series, the attention mechanism between the time periods is established and the global time series information is obtained to capture the global correlation. The step S5 specifically includes the following steps.

[0096] Step S5.1, for the divided new time series X∈R P×S×d , first encoding in the feature dimension d to d m ', the mapped data is:

[0097]

[0098] wherein, d m ' = Sxd.

[0099] Step S5.2, self-attention operation is performed between the time periods, thereby modeling the airport passenger flow data based on the correlation between the time periods. Including by linear mapping of X inter to obtain:

[0100]

[0101] Thus, three variables, Q inter , K inter , V interQuery, Key, Value represent the Query vector between time periods, the Key vector, and the Value vector, respectively. The Query vector is used to measure the relevance to each Key vector, the Key vector is used to build the calculation basis of the attention score, and the Value vector is used to perform weighted summation according to the attention score to generate the final attention output.

[0102] Step S5.3, obtain the attention information between time periods:

[0103]

[0104] Step S5.4, perform steps S5.1-5.3 between all time periods of the divided new time series to obtain the attention information between time periods as global time series information.

[0105] Step S6, after the above operations, the local time series information including the intra-period self-attention information and the global time series information including the inter-period self-attention information are obtained, which are fused with the Granger causality inference results obtained in the foregoing to achieve a weighted effect, and the prediction value of the airport passenger flow is obtained.

[0106] In this step, the model obtains the prediction value of the airport passenger flow by capturing the local details and global correlations of the time series data and fusing the strength information of the causal coupling.

[0107] Step S7, calculate the error between the obtained prediction value and the true value of the airport passenger flow data obtained in step S0, and update all weight parameters in the neural network using the gradient descent method.

[0108] Step S8, compare the obtained error with a predetermined threshold value, and compare the number of iterations with a maximum number of iterations, when one of the two conditions that the error is less than the predetermined threshold value and the number of iterations reaches the maximum number of iterations is satisfied, the training model ends, and a trained airport passenger flow prediction model is obtained, otherwise, return to step S3 to re-divide the time series in different scales S.

[0109] Step S9, input the measured real-time data to the obtained trained airport passenger flow prediction model to obtain airport passenger flow prediction information for airport resource scheduling and allocation. The real-time data is real-time obtained data and information in the airport, which can include passenger flow data, flight schedule, carbon dioxide concentration, PM2.5 concentration, temperature, humidity, etc.

[0110] In various scale airports, the airport passenger flow multi-factor fusion prediction method based on causal inference provided by the above embodiment of the application can be used to predict airport passenger flow. According to historical passenger flow data, flight schedules, carbon dioxide concentration, PM2.5 concentration, temperature, and humidity information, after model prediction, the airport passenger flow value at a future time can be obtained, thereby providing advanced data support for resource redistribution and traffic scheduling of the airport.

[0111] All the optional technical solutions described above can be combined to form optional embodiments of the application, and will not be described one by one here.

[0112] It should be understood that the size of the serial number of each step in the above embodiment does not mean the order of execution. The execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the application.

[0113] The above description is only a preferred specific implementation of the application, but the protection scope of the application is not limited to this. Any changes or replacements within the technical scope disclosed by the application can be easily thought of by those skilled in the art, and should be covered within the protection scope of the application.

Claims

1. An airport passenger flow multi-factor fusion prediction method based on causal inference, characterized in that, The method comprises the following steps: Step S0, obtaining input data information, including obtaining passenger flow data, flight schedule, carbon dioxide concentration, PM2.5 concentration, temperature and humidity information through a database and sensors, all information in the input data information is time series data; Step S1, based on the obtained input data information, performing Granger causality inference test on the airport passenger flow, constructing a multi-factor causal correlation, and obtaining input features for predicting the airport passenger flow; Step S2, determining whether the flight schedule, carbon dioxide concentration, PM2.5 concentration, temperature and humidity information in the input data information obtained in step S0 have been subjected to Granger test and analysis, if yes, taking all the obtained input features as the Granger causality inference result of the multi-factor, and performing the next step, otherwise, returning to step S1 to perform Granger causality inference test on the remaining influencing factors in the input data information; Step S3, dividing the time series of the input features to obtain new time series; Step S4, extracting features in each time period in the obtained new time series, establishing an attention mechanism in the time period and obtaining local time series information; Step S5, extracting features between time periods in the obtained new time series, establishing an attention mechanism between time periods and obtaining global time series information to capture global correlation; Step S6, fusing the obtained local time series information and global time series information with the Granger causality inference result of step S2 to realize a weighted effect, and obtaining a predicted value of the airport passenger flow; Step S7, calculating the error between the obtained predicted value and the true value of the passenger flow data obtained in step S0, and updating all weight parameters in the neural network by using gradient descent method; Step S8, comparing the obtained error with a predetermined threshold value, and comparing the number of iterations with a maximum number of iterations, when one of the two conditions that the error is less than the predetermined threshold value and the number of iterations reaches the maximum number of iterations is satisfied, the training model ends, and a trained airport passenger flow prediction model is obtained, otherwise, returning to step S3 to redivide the time series in different scales S; Step S9, inputting the measured real-time data into the obtained trained airport passenger flow prediction model to obtain airport passenger flow prediction information, which is used for airport resource scheduling and allocation.

2. The airport passenger flow multi-factor fusion prediction method based on causal inference according to claim 1, characterized in that, The step S1 specifically comprises the following steps: Step S1.1, selecting features from the flight schedule, carbon dioxide concentration, PM2.5 concentration, temperature and humidity information in the obtained input data information, and constructing a multivariate regression model of the airport passenger flow as follows: where Y t represents the airport passenger flow at time t, represents the value of the feature at time t-q, β p ,η t ,γ q represents the regression parameter and is trained from data, p represents the variable at time p, and q represents the variable at time q; Step S1.2, performing joint regression analysis by adding past values of the influencing factors, and constructing a univariate regression model of the airport passenger flow as follows: Y t = a0+ a1Y t-1 +…+ a p Y t-p + ε t where a p , e t denote regression parameters and are trained from data; Step S1.3, F-test is performed on the established multivariate regression model and univariate regression model of airport passenger flow to determine whether the past values of the features X t can significantly improve the prediction ability of the airport passenger flow Y t at time t.

3. The airport passenger flow multi-factor fusion prediction method based on causal inference according to claim 2, characterized in that, The step S1.3 specifically comprises the following steps: Step S1.3.1, performing F test on the established multivariate regression model and univariate regression model of the airport passenger flow to obtain a test value as follows: where RSS r is the residual sum of squares of the univariate autoregressive model, RSS ur is the residual sum of squares of the multivariate regression model, q is the number of lag terms introduced, n is the sample size, and k is the number of explanatory variables in the model. Step S1.3.2, analyzing the result obtained in step S1.3.1, if the obtained test value F is greater than the preset critical value, it indicates that the current feature X t Granger causes Y t Taking the current feature as the input feature for predicting the airport passenger flow, otherwise, it indicates that the feature has no effect on the prediction, and it is deleted.

4. The airport passenger flow multi-factor fusion prediction method based on causal inference according to claim 3, characterized in that, Step S3 specifically comprises: The time series data X e R H×d As input, where H denotes the length of the time series, d denotes the dimension of the feature, R H×d denotes a set of time series data, data partitioning is performed using a scale of length S, resulting in a new set of time series: (X1,...,X i ...,X P )∈R P×S×d wherein X i represents the input data of the i-th time period and X i ∈R S×d , P represents the number of time periods after the time series data is divided, each time period as a new unit.

5. The airport passenger flow multi-factor fusion prediction method based on causal inference according to claim 4, characterized in that, Step S4 specifically comprises the following steps: Step S4.1, encode the input data X i ∈R P in the feature dimension d for the i-th time segment in the divided new time series (X i 1,..., X S×d N ), obtaining: wherein, represents the new input data in the i-th time period after encoding on the feature dimension d, d m represents encoding the feature dimension d into d m ; Step S4.2, performing a linear transformation on to obtain a variable in the attention mechanism: wherein, represents a key vector over a time period, represents a value vector over a time period; Step S4.3, using trainable variables Merging the previous and next timing information in the time period, the input data X of the i-th time period is calculated i Attention information in: Step S4.4, performing steps S4.1-S4.3 on the input data of each time segment in the divided new time series, and the obtained attention information is the local temporal information.

6. The airport passenger flow multi-factor fusion prediction method based on causal inference according to claim 5, characterized in that, Step S5 specifically comprises the following steps: Step S5.1, encoding the divided new time series X ∈ R P×S×d , obtaining the mapped data as: ′ m , obtaining the mapped data as: wherein d ′ m = S x d; Step S5.2, performing a self-attention operation between time periods, modeling the airport passenger flow data based on the correlation between time periods, by applying a linear mapping of X inter to the variables between time periods: wherein Q inter represents a query vector between time periods, K inter represents a key vector between time periods, V inter represents a value vector between time periods; Step S5.3, obtaining the attention information between time segments: Step S5.4, performing steps S5.1-5.3 between all time segments of the divided new time series, and the obtained attention information between time segments is the global temporal information.

Images