Adaptive Algorithm to Improve Trajectory Prediction Accuracy of Climbing Aircraft

Abstract

Aircraft climb trajectories are difficult to predict, and large errors in these predictions reduce the potential operational benefits of some advanced concepts in the Next Generation Air Transportation System. An algorithm that dynamically adjusts modeled aircraft weights based on observed track data to improve the accuracy of trajectory predictions for climbing flights has been developed. In real-time evaluation with actual Fort Worth Center traffic, the algorithm decreased the altitude root-mean-square error by about 20%. It also reduced the root-mean-square error of predicted time at top of climb by the same amount.

Top-of-Climb Matching Method for Reducing Aircraft Trajectory Prediction Errors

“…The trajectory predictor evaluated in this study is the Center/TRACON Automation System (CTAS) Trajectory Synthesizer (TS) [23] that was analyzed in prior work [6][7][8][9][11][12][13][14][15]. CTAS is a real-time research prototype system developed at NASA that includes mature capabilities for 4-D trajectory prediction, conflict detection, conflict resolution, and other functions [24].…”

“…The adapted trajectory predictions more closely matched actual flight paths. Overall, the adaptive weight algorithm reduced both altitude and TOC time root mean square errors (RMSE) by about 20% for a data set comprised of about 400 actual Fort Worth Center climbing departures [11]. No aircraft types or climb profiles were intentionally included or excluded in that analysis.…”

“…Recent work on adaptive thrust [16][17] and adaptive weight [9,11] algorithms also demonstrated substantial improvements in trajectory prediction accuracy for actual climbing flights. Both algorithms only used the radar track and atmospheric forecast data available in current operations to dynamically adjust the modeled aircraft thrust or weight, respectively, on a per-flight basis.…”

“…Compared to methods that utilize the same adjusted aircraft performance model for all flights of the same aircraft type [12][13]19], the TOC-matching method proposed in this study is more effective because it is applied on a per-flight basis like the adaptive weight algorithm [9,11]. While the adaptive weight algorithm studies demonstrated the improvement in climb trajectory prediction accuracy that can be achieved using currently available data, this paper on the TOC-matching method demonstrates what may be possible in the near future.…”

“…These maneuvers increase both fuel burn and communication frequency congestion and could be avoided if the trajectory predictions for climbing flights were more accurate (especially in the vertical dimension) [10]. The TOC-matching method does not focus on improving horizontal trajectory prediction accuracy for climbing flights because only about one percent have horizontal errors that exceed the 5-nautical mile (nmi) horizontal separation standard [9,11] Researchers have improved climb trajectory prediction accuracy by adjusting aircraft performance models based on publicly available flight manuals [12] and analyzing trajectory errors [13]. The authors of the first study utilized model parameter identification to determine the combinations of thrust and drag that best modeled the time-to-TOC data specified in the Boeing 737-400 flight manual.…”