Cascading Delay Risk of Airline Workforce Deployments with Crew Pairing and Schedule Optimization

Area of Science:

• Operations Research
• Artificial Intelligence
• Aviation Management

Background:

• Flight disruptions, including delays and cancellations, are often caused by preceding flight arrival delays.
• Existing methods for assigning buffer times and reserve crews rely on historical delay data but may not capture complex causal factors.
• Accurate prediction of flight arrival delays is crucial for mitigating disruptions and optimizing crew scheduling.

Purpose of the Study:

• To develop a novel forecasting approach for predicting flight arrival delays.
• To determine optimal buffer times between connected flights.
• To establish a dynamic reserve crew strategy for efficient crew pairing and cost reduction.

Main Methods:

• Utilized a cascade neural network incorporating a massive amount of historical flight arrival and departure data.
• Incorporated learning ability within the network to uncover complex, non-linear relationships in the data.
• Developed a dynamic reserve crew strategy based on predicted flight arrival delays.

Main Results:

• Predicting flight departure delay as an input significantly increased the accuracy of flight arrival delay prediction.
• The proposed dynamic reserve crew strategy effectively reduced total crew costs.
• The approach demonstrated enhanced airline schedule stability through improved disruption mitigation.

Conclusions:

• Cascade neural networks offer a powerful tool for accurate flight delay prediction by learning intricate data relationships.
• The dynamic reserve crew strategy provides a cost-effective solution for airlines to manage disruptions.
• This big data-driven approach enhances operational efficiency and financial performance in the airline industry.