Large-scale competition civil aviation transportation flow prediction model and method thereof

Abstract

The invention relates to the technical field of civil aviation transportation, and discloses a large-scale competition civil aviation transportation flow prediction model and a method thereof, and the model comprises a feature extraction module which is used for extracting passenger flow volume change trends and hot air routes in different time periods from historical civil aviation transportation data, and extracting related features of the civil aviation transportation flow in combination with economic and traffic data; according to the method, through the strong processing capability of the fusion model to the time sequence and the consideration of the model to various factors, the complex change rule of the civil aviation transportation flow in time and space during the large competition period can be fully captured, and compared with a traditional single model, the prediction accuracy is remarkably improved; a mean square error is adopted as a loss function and a stochastic gradient descent algorithm is adopted for optimization, the learning rate and bandwidth parameters are continuously adjusted in the training process, the model achieves a good effect through multiple iterations, the fitting ability and generalization ability of the model to data are enhanced, and the prediction precision and stability are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of civil aviation transportation, and particularly to a prediction model and method for civil aviation transportation flow during large-scale events. Background Technique

[0002] With the global development and sports culture exchange activities, there are various large-scale events. And large-scale events often gather numerous audiences, athletes and staff from home and abroad, who flow between different locations, thus forming a huge tide of personnel flow. Among various personnel transportation methods, civil aviation transportation, with its convenient and fast characteristics, shows unique advantages in long-distance travel and becomes an important undertaker of personnel transportation tasks, undertaking a considerable proportion of personnel transportation tasks.

[0003] Accurately predicting the civil aviation transportation flow during large-scale events is of great significance for airlines to reasonably allocate flight resources, for airports to optimize operation management, and for relevant departments to formulate traffic guidance strategies. However, there are currently many problems in such predictions: First, traditional prediction models mostly rely on historical data for simple trend analysis, ignoring the complex impacts of special factors such as the scale of the event, the location, time, and type of the event on passengers' travel decisions; Second, existing methods fail to effectively integrate multi-source data and cannot comprehensively and accurately present the dynamic change characteristics of the civil aviation transportation flow during this period. Summary of the Invention

[0004] The purpose of the present invention is to provide a prediction model and method for civil aviation transportation flow during large-scale events to solve the problems raised in the above background technique.

[0005] To achieve the above purpose, the present invention provides the following technical solution: A prediction model for civil aviation transportation flow during large-scale events, the model includes:

[0006] A data collection module, which collects relevant data of large-scale events and uniformly integrates the basic information of the event, flight aircraft data, tickets, and social media data;

[0007] A data analysis module, which analyzes data from different sources and performs cleaning, transformation, and normalization processing on the data;

[0008] A feature extraction module, which extracts the changing trends of passenger flow and popular routes in different time periods from historical civil aviation transportation data, and combines economic and traffic data to extract relevant features related to the civil aviation transportation flow;

[0009] A model construction module, which constructs a model framework, processes time series data, captures the dependency relationship of the civil aviation transportation flow in the time dimension, and at the same time combines the relevant features of each piece of information to jointly construct a prediction model;

[0010] The model evaluation and prediction module optimizes the processed model and applies it to the large-scale event data to be detected, outputs the predicted civil aviation transportation flow results, and presents them to the user in the form of charts.

[0011] Preferably, the data acquisition module includes:

[0012] Basic event information: Extract relevant text information from the official website through web crawler technology, extract date information, and obtain the list of athletes on the registration page.

[0013] Flight operation information: Based on the airline website and booking platform, obtain the content of the operation plan according to relevant procedures, and conduct relevant statistics on the relevant data and technology of the aircraft.

[0014] Ticket information: Establish a data cooperation relationship with the official event ticketing system or ticketing agency platform, obtain the sales quantity data in the way stipulated by the agreement, and obtain the sales report of the platform irregularly.

[0015] Social media data information: While scraping the post content, obtain the interaction data of the number of likes, comments, and reposts corresponding to each post to measure the popularity and spread of the post.

[0016] Preferably, the data analysis module, based on time series analysis, assumes that R t is the sales quantity of tickets at time t, combines the correlation between various information, and its calculation formula is;

[0017]

[0018] where and R t are model parameters, ∈ t is the coefficient sequence between information, w and m are the autoregressive order and moving average order respectively. After determining by fitting the historical sales volume, the ticket quantity at future times is predicted.

[0019] Preferably, the feature extraction module extracts the basic event information feature A, flight operation information feature B, ticket information feature C, and social media data feature D respectively, where;

[0020] The calculation formula for the basic event information feature is;

[0021] The calculation formula for the flight operation information feature is: where t i is the number of seats of the i-th flight, and n is the total number of flights;

[0022] The calculation formula for the ticket information feature is:

[0023] The calculation formula for the characteristics of social media data is as follows: Where Nice, Remark, and Retweets are the number of likes, comments, and retweets of the i-th post respectively, and n is the total number of posts.

[0024] Preferably, the model construction module establishes a model according to the relevance between the object index and the past index, using historical data and combining various characteristic values. The formula set according to this is:

[0025]

[0026] Where i = A, B, C, D, which are the average values of the object index and the past index respectively, that is where n is the total number of object indexes.

[0027] Preferably, the model evaluation and prediction module inputs the divided data set into the constructed model for calculation, continuously adjusts the model parameters during the training process, and the optimization algorithm updates the model parameters in each iteration, making the loss function value gradually decrease, which is beneficial to improving the prediction accuracy. Its construction formula is;

[0028] F t = αX t +(1 - α)F t-1 ;

[0029] Where F t represents the smoothing index in the previous period of time. It is divided into the subsequent middle period and the later period according to time. X t represents the observed value of the object index in the t time period, and α is the smoothing coefficient, which is set between 0.2 and 0.9. According to this coefficient, it expands in a decreasing manner:

[0030] F t = αX t +(1 - α)F t-1 =......αX t +(1 - α) -1 X 1 ;

[0031] Preferably, the model evaluation and prediction module applies the model after the optimization algorithm to the large-scale event data to be predicted, and outputs the predicted civil aviation traffic flow result. The prediction result is presented to the user in the form of a chart or a data report.

[0032] A method for predicting the civil aviation traffic flow of a large-scale event, the prediction method includes:

[0033] S1: In the research on the civil aviation transportation flow of large-scale events, collect multi-source data, obtain the event holding time, location, expected scale and type, clarify the basic structure of the event, extract the number of flight takeoffs and landings and the passenger flow of each route during the same season and events of similar scale in the past five years, obtain the historical transportation situation, cities, surrounding highway and railway transport capacities, obtain the distribution and operation data of transportation hubs, and integrate data from different sources;

[0034] S2: Clean the recorded data after integration. Based on the time dimension, predict the total number of tickets in the future, analyze the internal relationship between civil aviation transportation and event tickets, and evaluate its development trend;

[0035] S3: Analyze the historical civil aviation transportation data, extract the passenger flow change trends within different time windows and at different times during past similar events, combine algorithms with the extracted features, calculate the distance between the venue and the airport at the event location, the traffic congestion index, and the traffic accessibility between the main passenger sources and the event location;

[0036] S4: Use the training set to train the constructed model. With the mean square error as the loss function, during training, continuously adjust the model learning rate and model parameters, and through multiple iterations, train the model;

[0037] S5: Input the processed data into the trained model. The prediction output module generates the predicted data of the passenger flow, number of times, and passenger source - destination distribution of each route during each period of the event, and presents the daily passenger flow curve and the bar chart of the passenger flow of each route in the form of a chart.

[0038] Compared with the prior art, the beneficial effects of the present invention are:

[0039] In the present invention, by integrating the powerful processing ability of the model for time series and the consideration of various factors by the model, it is possible to fully capture the complex change laws of the civil aviation transportation flow in terms of time and space during large-scale events. Compared with traditional single models, the prediction accuracy is significantly improved. And by using the mean square error as the loss function and the stochastic gradient descent algorithm for optimization, continuously adjust the learning rate and bandwidth parameters during the training process, and through multiple iterations, the model reaches a better effect, enhancing the model's fitting ability and generalization ability for data, and improving the prediction accuracy and stability.

[0040] In the present invention, through accurate prediction and reasonable operation decisions, it is possible to optimize the allocation of civil aviation transportation resources, improve the operation efficiency of the entire civil aviation transportation industry during large-scale events, reduce resource waste, achieve the sustainable development of the industry, ensure that passengers can travel smoothly and conveniently during the event, reduce the inconvenience brought to passengers by problems such as unreasonable flight arrangements and inadequate airport services, enhance the passengers' travel experience, and thus enhance the image and reputation of the civil aviation transportation industry. Description of the Drawings

[0041] Figure 1 It is a step diagram of a civil aviation transportation flow prediction model for a large-scale event;

[0042] Figure 2 It is a construction diagram of a feature extraction module of a civil aviation transportation flow prediction model for a large-scale event. Specific implementation manners

[0043] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0044] Please refer to Figure 1 - Figure 2 , the present invention provides a technical solution: a civil aviation transportation flow prediction model for a large-scale event, and the model includes:

[0045] A data collection module, which collects relevant data of a large-scale event and uniformly integrates the basic information of the event, flight data, tickets, and social media data;

[0046] A data analysis module, which analyzes data from different sources and performs cleaning, transformation, and normalization processing on the data;

[0047] A feature extraction module, which extracts the changing trends of passenger flow and popular routes in different time periods from historical civil aviation transportation data, combines economic and traffic data, and extracts relevant features related to the civil aviation transportation flow;

[0048] A model construction module, which constructs a model framework, processes time series data, captures the dependence relationship of the civil aviation transportation flow in the time dimension, and combines the relevant features of each piece of information to jointly construct a prediction model;

[0049] A model evaluation and prediction module, which optimizes the processed model, applies it to the data of the large-scale event to be detected, and outputs the predicted civil aviation transportation flow results, presenting them to the user in the form of charts.

[0050] The following further illustrates this solution in conjunction with Embodiment 1 to Embodiment 3:

[0051] Embodiment 1:

[0052] The data collection module of the present invention includes:

[0053] The basic information of the event, through web crawler technology, extracts relevant text information from the official website, extracts date information, and obtains the list of athletes on the registration page;

[0054] Flight operation information, based on airline websites and booking platforms, obtains the content of the operation plan according to relevant processes, and conducts relevant statistics on the relevant data and technologies of the aircraft;

[0055] Ticket information, establishes a data cooperation relationship with the official ticketing system of the event or the ticketing agency platform, obtains the sales quantity data in the way stipulated by the agreement, and obtains the sales report of the platform irregularly;

[0056] Social media data information, while capturing the content of posts, obtains the interaction data of the number of likes, comments, and forwards corresponding to each post to measure the popularity and spread of the post;

[0057] Data analysis module, based on time series analysis, assuming R t is the sales quantity of tickets at time t, combined with the correlation between various information, and its calculation formula is;

[0058]

[0059] where and R t are model parameters, ∈ t is the coefficient sequence between information, w and m are the autoregressive order and the moving average order respectively. After determining by fitting the historical sales volume, the ticket quantity at future times is predicted;

[0060] Feature extraction module, extracts the basic information features A of the event, the flight operation information features B, the ticket information features C, and the social media data features D respectively, where;

[0061] The calculation formula for the basic information features of the event is;

[0062] The calculation formula for the flight operation information features is: where t i is the number of seats of the i-th flight, and n is the total number of flights;

[0063] The calculation formula for the ticket information features is:

[0064] The calculation formula for the social media data features is: where Nice, Remark, and Retweets are the number of likes, comments, and forwards of the i-th post respectively, and n is the total number of posts;

[0065] First, measure the popularity of the post by capturing the content of social media posts and interaction data such as likes, comments, and forwards. The data analysis module is based on time series analysis and uses the formula

[0066]

[0067] Among them and R t are model parameters, ∈ t is the coefficient sequence between information, w and m are the autoregressive order and the moving average order respectively. First, the parameters are determined by fitting the historical sales volume, and then the future ticket quantity is predicted. At the same time, the feature extraction module calculates the basic information features of the event (the ratio of the number of athletes to the expected number of spectators), the flight operation information features (the ratio of the total number of flight seats to the total number), the ticket information features (the ratio of the total number of tickets to the total number of ticket sales dates), and the social media data features (the ratio of the total number of post interactions to the total number) respectively, providing multi-dimensional data support for the prediction and making the prediction more comprehensive and accurate.

[0068] Example 2:

[0069] The model construction module of the present invention, according to the relevance between the object index and the past index, can use historical data and combine each feature value to establish a model. The formula set according to this is:

[0070]

[0071] where i = A, B, C, D, are the average values of the object index and the past index respectively, that is where n is the total number of objects and indexes;

[0072] This formula measures the correlation degree between the object index and the past index by calculating the difference relationship between each index data and the average value, providing a basis for model construction, and can be used to analyze the correlation relationship between different features (such as event basic information, flight operation information, ticket information, social media data, etc.) and related indexes, helping with prediction and analysis work.

[0073] Example 3:

[0074] The model evaluation and prediction module of the present invention inputs the divided data set into the constructed model for calculation, continuously adjusts the model parameters during the training process, and the optimization algorithm updates the model parameters in each iteration, making the loss function value gradually decrease, which is beneficial to improving the prediction accuracy. Its construction formula is;

[0075] F t = αX t +(1 - α)F t-1 ;

[0076] where F t represents the smoothing index in the previous period of time, which is divided into the subsequent middle period and the later period according to time, X t represents the observed value of the object index in the t time period, α is the smoothing coefficient, which is set between 0.2 and 0.9, and is expanded in decreasing order according to this coefficient:

[0077] F t = αX t +(1 - α)F t-1 =......αX t +(1 - α) -1 X 1 ;

[0078] The model evaluation and prediction module applies the model after the optimization algorithm to the large-scale event data to be predicted, outputs the predicted civil aviation traffic flow results, and presents the prediction results to the user in the form of charts or data reports;

[0079] First, input the divided data set into the constructed model for calculation. During the training process, continuously adjust the model parameters, update the parameters in each iteration through the optimization algorithm, and gradually reduce the loss function value to improve the prediction accuracy. Its construction formula is F t = αX t +(1 - α)F t-1 ;

[0080] where F t represents the smoothing index in the previous period of time. According to the time division, there are subsequent middle and late periods. It is the observed value of the object index in the time period. α is the smoothing coefficient (set between 0.2 and 0.9), and it expands in a decreasing manner according to this coefficient. Finally, apply the optimized model to the large-scale event data to be predicted, output the prediction results of the civil aviation traffic flow, and present them to the user in an intuitive chart or data report form for the user to understand and use.

[0081] In the present invention, in the research on the prediction of civil aviation transportation flow for large-scale events, multi-source data is comprehensively collected, including the event holding time, location, expected scale and type. The number of flight takeoffs and landings and the passenger flow of each route during the same season and events of similar scale in the past five years are obtained from the operation databases of airports and airlines. The GDP, population, and per capita disposable income of the host city and the main source regions are obtained from the websites of government statistical departments. The highway and railway transport capacities of the city and its surrounding areas, as well as the distribution and operation data of transportation hubs, are obtained from the transportation department. Then, the data from different sources is integrated, the historical civil aviation transportation data is cleaned, abnormal records of passenger flow caused by errors are removed, and it is unified into a standard format. Based on the time dimension and the relationships of various information, the total number of future tickets is predicted. The historical civil aviation transportation data is deeply analyzed, the changing trends of passenger flow in different time windows of past similar events are extracted, and combined with algorithms and the extracted features, the distance between the venue and the airport at the host location, the traffic congestion index, and the traffic accessibility between the main source regions and the host location are further obtained. The constructed model is trained using the training set, the mean square error is used as the loss function, and the stochastic gradient descent algorithm is used for optimization. During the training, the learning rate and bandwidth parameters are continuously adjusted. After multiple iterations, when the mean square error of the validation set converges to a small value, it indicates that the model training effect is good. The processed data is input into the trained model, and the prediction output module gives the passenger flow of each route, the number of flight takeoffs and landings, and the origin-destination distribution during each period of the event, and the results are presented in the form of charts, such as the daily passenger flow change curve and the column chart of the passenger flow of each route, providing an intuitive basis for the operation decisions of airlines and airports. Airlines reasonably adjust the flight frequencies and aircraft types of popular routes according to the predicted passenger flow of each route, and airports arrange terminal service personnel and allocate baggage handling equipment resources in advance according to the predicted passenger throughput.

[0082] The present invention also includes a method for predicting the civil aviation transportation flow of large-scale events. The prediction method includes:

[0083] Step 1: In the research on the civil aviation transportation flow of large-scale events, multi-source data is collected, the event holding time, location, expected scale and type are obtained, the basic structure of the event is clarified, the number of flight takeoffs and landings and the passenger flow of each route during the same season and events of similar scale in the past five years are extracted, the historical transportation situation, the city, the surrounding highways and railway transport capacities are obtained, the distribution and operation data of transportation hubs are obtained, and the data from different sources is integrated;

[0084] Step 2: Clean the recorded data after integration, predict the total number of future tickets based on the time dimension, analyze the internal relationship between civil aviation transportation and event tickets, and evaluate its development trend;

[0085] Step 3: Analyze the historical civil aviation transportation data, extract the passenger flow change trends at different time windows and different times during past similar events, and combine the algorithm with the extracted features to calculate the distance between the venue and the airport at the hosting location, the traffic congestion index, and the traffic accessibility between the main source of passengers and the hosting location;

[0086] Step 4: Use the training set to train the constructed model. With the mean squared error as the loss function, during training, continuously adjust the model learning rate and model parameters, and through multiple iterations, train the model;

[0087] Step 5: Input the processed data into the trained model, and the prediction output module generates prediction data on the passenger flow, frequency, and origin-destination distribution of each route during each period of the event, and presents the daily passenger flow curve and the bar chart of the passenger flow of each route in the form of a graph.

[0088] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A civil aviation transport flow prediction model for large-scale events, characterized by: The model includes: The data collection module collects relevant data of large-scale events and integrates basic information of the events, flight data, ticket data and social media data; The data analysis module analyzes data from different sources and cleans, transforms and normalizes the data; The feature extraction module extracts the passenger flow change trends and popular routes in different time periods from historical civil aviation transportation data, and extracts the relevant features of civil aviation transportation flow by combining economic and traffic data; Model building module: builds the model framework, processes time series data, captures the dependencies of civil aviation transport flows in the time dimension, and combines the relevant features of each information to jointly build a prediction model; The model evaluation and prediction module optimizes the processed model, applies it to the large-scale event data to be tested, outputs the predicted civil aviation transport flow results, and presents them to the user in the form of charts.

2. A large-scale event civil aviation transport flow prediction model according to claim 1, characterized in that: The data acquisition module comprises: Basic information of the event: using crawler technology, extract relevant text information from the official website, extract date information, and obtain the list of athletes on the registration page; Flight operation information, based on the airline website and booking platform, obtain the operation plan content according to the relevant process, and make relevant statistics on the relevant data and technology of the aircraft; Ticket information: establish data cooperation with the official ticketing system or ticketing agent platform of the event, obtain sales quantity data in accordance with the method specified in the agreement, and obtain sales reports from the platform from time to time; Social media data information, while crawling the post content, obtains the interactive data of the number of likes, comments and reposts corresponding to each post to measure the popularity and dissemination range of the post.

3. A large-scale event civil aviation transport flow prediction model according to claim 1, characterized in that: The data analysis module is based on time series analysis and assumes that R t is the number of tickets sold at time t. Combined with the correlation between various information, the calculation formula is: in and R t is the model parameter, ∈ t is the coefficient sequence between information, w and m are the autoregressive order and the moving average order respectively. After fitting the historical sales, the number of tickets at future times is predicted.

4. A large-scale event civil aviation transport flow prediction model according to claim 1, characterized in that: The feature extraction module extracts the event basic information feature A, the flight operation information feature B, the ticket information feature C and the social media data feature D respectively, wherein; The calculation formula for the basic information characteristics of the event is: The calculation formula for flight operation information characteristics is: where t i is the number of seats on the ith flight, and n is the total number of flights; The calculation formula for ticket information features is: The formula for calculating social media data features is: Among them, Nice, Remark and Retweets are the number of likes, comments and reposts of the i-th post respectively, and n is the total number of posts.

5. The large-scale event civil aviation transport flow prediction model according to claim 1 is characterized in that: The model building module can build a model based on the correlation between the object index and the past index by using historical data and combining various feature values. The formula according to this setting is: Where i = A, B, C, D, are the average values ​​of the target index and the past index, that is, Where n is the total number of objects and indicators.

6. A large-scale event civil aviation transport flow prediction model according to claim 1, characterized in that: The model evaluation prediction module inputs the divided data set into the constructed model for calculation, continuously adjusts the model parameters during the training process, and the optimization algorithm updates the model parameters in each iteration so that the loss function value gradually decreases, which is conducive to improving the prediction accuracy. The construction formula is: F t =αX t +(1-α)F t-1 ; where F t It represents the smoothing index in the early stage of time, which is divided into the subsequent middle stage and the late stage according to the time division. t It is expressed as the observed value of the target indicator in the t time period, α is the smoothing coefficient, which is set between 0.2-0.

9. According to this coefficient, it is expanded in decreasing order: F t =αX t +(1-α)F t-1 =......αX t +(1-a) -1 X1.

7. A large-scale event civil aviation transport flow prediction model according to claim 6, characterized in that: The model evaluation and prediction module applies the model after the optimization algorithm to the large-scale event data to be predicted, outputs the predicted civil aviation transportation flow results, and presents the prediction results to the user in the form of charts or data reports.

8. A method for predicting civil aviation transport flows for large-scale events, referring to a prediction model for civil aviation transport flows for large-scale events as described in any one of claims 1 to 7, the prediction method comprising: S1: In the study of civil aviation transport flows for large-scale events, collect multi-source data, obtain the time, location, expected scale and type of the event, clarify the basic structure of the event, extract the flight take-offs and landings and passenger flow of each route during the same season and similar-scale events in the past five years, obtain historical transportation conditions, cities, surrounding roads and railway capacity, obtain distribution and operation data of transportation hubs, and integrate data from different sources; S2: After cleaning and integrating the recorded data, based on the time dimension, predict the total number of tickets in the future, analyze the intrinsic connection between civil aviation transportation and event ticketing, and evaluate their development trends; S3: Analyze historical civil aviation transportation data, extract passenger flow trends in different time windows and at different times during similar events in the past, combine algorithms with extracted features, calculate the distance between the venue and the airport, traffic congestion index, and traffic accessibility between the main passenger source areas and the venue; S4: Use the training set to train the constructed model, using the mean square error as the loss function. During training, continuously adjust the model learning rate and model parameters, and train the model after multiple iterations; S5: The processed data is input into the trained model, and the prediction output module generates the predicted data of passenger flow, frequency and source-destination distribution of each route in each period during the event, and presents the daily passenger flow curve and the passenger flow bar chart of each route in a chart.

Images