2026 年 69 巻 3 号 p. 61-74
As air traffic systems transition towards trajectory-based operations with fewer constraints on flight planning, optimizing trajectories in flight plan generation holds promise for greater efficiency. To harness this potential, the optimizer necessitates an aircraft performance model with highly accurate fuel flow rate predictions, capable of accounting for individual aircraft performance variations over time due to daily wear-and-tear of flight operations and maintenance. This study proposes a method to construct tailored performance models using airline-acquired flight data. We demonstrate the feasibility of simultaneously estimating aerodynamic and fuel models, traditionally regarded as challenging, by incorporating vertical velocity due to climb and descent, and inertial drag due to acceleration and deceleration. This approach reduces estimation bias, enhancing the model’s applicability across a wider flight envelope including climb and descent phases, thus facilitating more comprehensive trajectory optimization solutions. Hold-out validation confirms the model’s accuracy, with mean fuel flow rate error below 1%. Furthermore, this method’s ability to extract individual aircraft performance from flight data suggests potential applications in maintenance-related performance monitoring.
