Development of a Metric Concept that Differentiates Between Normal and Abnormal Operational Aviation Data

Area of Science:

• Aviation safety
• Data science
• Risk management

Background:

• Airlines possess vast operational data crucial for safety performance monitoring.
• Current safety performance indicators rely on exceedance-based methods, lacking real-time operational risk detail.
• Regulators mandate data-driven safety performance demonstration.

Purpose of the Study:

• To develop novel metrics complementing existing exceedance-based safety approaches.
• To create aggregate construct variables for differentiating normal and abnormal landings.
• To identify temporal sequence patterns indicative of landing types.

Main Methods:

• Development of two construct variables: row_mean (aggregate) and row_sequence (temporal patterns).
• Application of statistical and visual tests to compare landing data series.
• Validation using time series k-means cluster analysis.

Main Results:

• Composite constructs effectively differentiate normal from abnormal landings.
• The metrics capture time-varying variable importance in the final 300 seconds before touchdown.
• Successful differentiation verified through cluster analysis.

Conclusions:

• The proposed metrics offer a valuable complement to traditional safety performance indicators.
• These data-driven approaches enhance real-time operational risk assessment.
• Further research into these complementary methods is recommended for aviation safety.