Depth flight pricing method based on deep learning

Abstract

The invention relates to a stop flight pricing method based on deep learning. The method comprises the following steps: 1) fitting a ticket price-passenger flow relation curve by using a deep learning method through historical flight data and external data; and 2) the fitting curve in the step 1) is used to optimize the flight income by using a doliolaria group heuristic swarm intelligence algorithm. Through a hierarchical structure of the neural network, the method can automatically learn related knowledge such as selection behaviors of passengers, fluctuation of market demands, competition between airlines and other complex modes from historical data, provides more accurate support for a pricing model, can process a large amount of input variables and massive historical data, and improves the accuracy of pricing. When the demand of the stopped flight is predicted, the deep learning model can comprehensively consider a plurality of factors and effectively identify key factors influencing ticket price changes in the training process.

Description

Technical Field

[0001] The present invention relates to the field of intersectional technology of artificial intelligence and civil aviation, and specifically to a stopover flight pricing method based on deep learning. Background Art

[0002] A stopover flight is a flight that requires a stopover at one or more transfer airports during the flight, and passengers take the same plane to the final destination after a period of stay at the airport. The pricing of stopover flights usually depends not only on the length of the flight and the starting and ending points, but also on multiple variables. For example, the passenger's transfer waiting time, flight schedule, seasonal demand, flight seat supply, and competition at transfer airports will all affect the setting of flight fares. What's more complicated is that the pricing of stopover flights also needs to take into account the fares of different segments and the pricing strategies between airlines, so it involves an optimization problem across airlines and segments. In addition, passenger selection behavior and fluctuations in flight demand will further increase the complexity of the pricing model. How to balance factors such as passenger flow demand, flight costs and market competition while ensuring profitability is a key difficulty in the pricing of stopover flights.

[0003] In recent years, deep learning has been widely used in the civil aviation field, especially in flight passenger flow forecasting. Flight passenger flow forecasting is an important basis for airlines to formulate fare strategies, dispatch flight resources, optimize route networks, and improve operational efficiency. Deep learning methods, especially deep neural networks, long short-term memory networks, and convolutional neural networks, can more accurately capture the laws of flight passenger flow changes through their powerful data processing capabilities and ability to learn complex patterns. In addition, deep learning is also widely used in flight delay prediction, flight scheduling optimization, passenger behavior analysis, and even aviation safety. It can optimize flight scheduling plans, reduce operating costs, and improve flight punctuality by learning from a large amount of flight operation data. In the fiercely competitive civil aviation field, the application of deep learning helps airlines better grasp market dynamics and formulate more flexible operating strategies.

[0004] Combining the pricing problem of stopover flights and the application of deep learning in the civil aviation field, compared with the traditional stopover flight pricing method based on manual key factor mining, it can be found that deep learning has considerable potential in solving the stopover flight pricing problem:

[0005] 1. The pricing of stopover flights involves complex patterns such as passengers' choice behavior, fluctuations in market demand, and competition among airlines. Traditional pricing models may find it difficult to capture the nonlinear relationship between these complex factors. Deep learning, through the hierarchical structure of neural networks, can automatically learn relevant knowledge from historical data and provide more accurate support for pricing models.

[0006] Second, deep learning can process a large number of input variables and massive amounts of historical data. When predicting the demand for stopover flights, deep learning can comprehensively consider multiple factors and effectively identify the key factors that affect ticket price changes during the training process, thereby optimizing pricing strategies. This has significant advantages over key variable identification based on expert systems.

[0007] 3. Deep learning can enhance the accuracy of the prediction model by integrating different external data sources, such as weather data, economic trends, social media information, etc. In the pricing of stopover flights, considering the different demand characteristics of different flight segments and transfer stations, deep learning can adaptively adjust the price strategy of each flight segment, thereby optimizing the overall pricing structure.

[0008] To this end, a stopover flight pricing method based on deep learning is designed to overcome the above problems. Summary of the invention

[0009] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a stopover flight pricing method based on deep learning that improves pricing accuracy and enhances the flexibility and profitability of airlines in market competition. The present invention maximizes the profits of airlines while meeting the travel needs of passengers and promotes the rational allocation of flight resources.

[0010] The present invention is implemented by the following technical solution: a stopover flight pricing method based on deep learning, the method comprising the following steps:

[0011] 1) Use deep learning methods to fit the “fare-passenger flow” relationship curve through historical flight data and external data;

[0012] 2) The fitting curve in step 1) is used to optimize the flight revenue using the salp swarm heuristic swarm intelligence algorithm.

[0013] Preferably, the step 1) is specifically as follows:

[0014] 1) Collect and clean the historical sales process data and result data of each flight;

[0015] 2) For each time point that appears in the historical flight data, relevant external data is collected from trusted channels, such as weather conditions, economic trends, social media information, etc., and the historical flight data and relevant external data are combined into time series data;

[0016] 3) Use the Transformer model to train the time series data obtained in step 2 to obtain the fitted "fare-passenger flow" relationship curve.

[0017] Preferably, the step 2) is specifically as follows:

[0018] 1) Randomly assign the ticket prices of each segment of the flight as the initial salp population;

[0019] 2) Use the formula Update the leader position, where represents the fare of the flight segment numbered j in the i-th particle, F j represents the current optimal solution position, c1 is the parameter for balancing development and exploration, t is the current iteration number, T is the maximum iteration number, c2 and c3 are random numbers between 0 and 1;

[0020] 3) Use the formula Update each follower's position;

[0021] 4) Repeat steps 2 and 3 until the maximum number of iterations is reached, and take the optimal solution at this time as the final pricing plan.

[0022] The beneficial effects of the present invention are as follows:

[0023] Compared with the prior art, the stopover flight pricing method based on deep learning provided by the present invention, by introducing deep learning into the stopover flight pricing problem, can achieve the following positive effects:

[0024] (1) Through the hierarchical structure of the neural network, this method can automatically learn relevant knowledge such as passengers' choice behavior, fluctuations in market demand, competition among airlines and other complex patterns from historical data, providing more accurate support for the pricing model.

[0025] (2) This method is able to process a large number of input variables and massive amounts of historical data. When predicting the demand for stopover flights, the deep learning model can comprehensively consider multiple factors and effectively identify the key factors that affect ticket price changes during the training process.

[0026] (3) This method can enhance the accuracy of the prediction model by integrating different external data sources, such as weather data, economic trends, social media information, etc. At the same time, according to the different demand characteristics of different flight segments and transit stations, deep learning can adaptively adjust the pricing strategy of each flight segment, thereby optimizing the overall pricing structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a schematic diagram of the “fare-passenger flow” relationship curve fitting process based on deep learning (using the Transformer model).

[0028] Figure 2 Schematic diagram of the pricing decision algorithm based on the salp swarm heuristic swarm intelligence algorithm. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to more clearly understand the objectives, technical solutions and advantages of the present invention, the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0030] In the description of the present invention, it is necessary to understand that the orientation or position relationship indicated by the terms such as "up", "down", "left", "right", "inside", "outside", "horizontal" and "vertical" is based on the orientation or position relationship shown in the accompanying drawings, and is only for the convenience of describing the present invention, rather than indicating or implying that the device or original referred to must have a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0031] The present invention will be described in detail below with reference to the accompanying drawings: Figure 1-2 As shown, a stopover flight pricing method based on deep learning comprises the following steps:

[0032] 1) Use deep learning methods to fit the “fare-passenger flow” relationship curve through historical flight data and external data;

[0033] 2) The fitting curve in step 1) is used to optimize the flight revenue using the salp swarm heuristic swarm intelligence algorithm.

[0034] Preferably, the step 1) is specifically as follows:

[0035] 1) Collect and clean the historical sales process data and result data of each flight;

[0036] 2) For each time point that appears in the historical flight data, relevant external data is collected from trusted channels, such as weather conditions, economic trends, social media information, etc., and the historical flight data and relevant external data are combined into time series data;

[0037] 3) Use the Transformer model to train the time series data obtained in step 2 to obtain the fitted "fare-passenger flow" relationship curve.

[0038] Preferably, the step 2) is specifically as follows:

[0039] 1) Randomly assign the ticket prices of each segment of the flight as the initial salp population;

[0040] 2) Use the formula Update the leader position, where represents the fare of the flight segment numbered j in the i-th particle, F j represents the current optimal solution position, c1 is the parameter for balancing development and exploration, t is the current iteration number, T is the maximum iteration number, c2 and c3 are random numbers between 0 and 1;

[0041] 3) Use the formula Update each follower's position;

[0042] 4) Repeat steps 2 and 3 until the maximum number of iterations is reached, and take the optimal solution at this time as the final pricing plan.

[0043] Example 1

[0044] Suppose an airline has a stopover flight from A to C via B, with 200 seats. The segments sold to the outside include AB, BC, and AC, and the optimization goal is the total revenue of the flight. First, the historical sales process data and sales result data of each route and flight mastered by the airline are used as historical data, and the meteorological data provided by the China Meteorological Administration and the Baidu search index of the route departure and destination on the day of the flight are used as external data. The two are integrated to form a time series vector data of flight passenger flow and external factors. The Transformer model is trained based on this data, and the external factor vector of the flight to be optimized is used as the input parameter to obtain the fare-passenger flow fitting curve of each segment of the flight to be predicted. Assume that the fare-passenger flow fitting curves of the three segments are:

[0045]

[0046]

[0047]

[0048] The above fitting curve is used to run the pricing decision algorithm based on the salp swarm heuristic swarm intelligence algorithm, setting the maximum number of iterations to 1000 and the number of populations to 50. The optimal pricing is: AB segment fare 625 yuan, BC segment fare 225 yuan, AC segment fare 725 yuan, and total revenue 149,375 yuan, which is consistent with the manual calculation results.

[0049] Anyone familiar with the technology may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by a person of ordinary skill in the art without departing from the spirit and technical concept disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A stopover flight pricing method based on deep learning, characterized in that: The method comprises the following steps: 1) Use deep learning methods to fit the "fare-passenger flow" relationship curve through historical flight data and external data; 2) The fitting curve in step 1) is used to optimize the flight revenue using the salp swarm heuristic swarm intelligence algorithm.

2. The stopover flight pricing method based on deep learning according to claim 1, characterized in that: The step 1) is specifically as follows: 1) Collect and clean the historical sales process data and result data of each flight; 2) For each time point that appears in the historical flight data, relevant external data is collected from trusted channels, and the historical flight data and relevant external data are combined into time series data; 3) Use the Transformer model to train the time series data obtained in step 2 to obtain the fitted "fare-passenger flow" relationship curve.

3. The stopover flight pricing method based on deep learning according to claim 1 or 2, characterized in that: The step 2) is specifically as follows: 1) Randomly assign the ticket prices of each segment of the flight as the initial salp population; 2) Use the formula Update the leader position, where represents the fare of the flight segment numbered j in the i-th particle, F j represents the current optimal solution position, c1 is the parameter for balancing development and exploration, t is the current iteration number, T is the maximum iteration number, c2 and c3 are random numbers between 0 and 1; 3) Use the formula Update each follower's position; 4) Repeat steps 2 and 3 until the maximum number of iterations is reached, and take the optimal solution at this time as the final pricing plan.

Images