Knowledge Transfer Framework for Airspace Situation Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current air traffic management systems face challenges in accurately evaluating sector situation due to limited training samples and the inability to utilize samples from non-target sectors, leading to over-burdened air traffic controllers and operational errors.
Innovation Solution
A sector situation evaluation framework based on knowledge transfer that mines knowledge from both target and non-target samples using multi-factor subset generation and multi-base evaluator construction, with sample transformation strategies to integrate evaluators effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional machine learning models are used for sector situation evaluation, then evaluation performance can be satisfactory with sufficiently large sample sets, but the sample collection becomes time-consuming and labor-intensive requiring ATM expert participation
Solution Approach 1:
The patent segments the evaluation task into multiple base evaluators, each responsible for specific situation factors. This segmentation allows parallel processing and reduces the time required for comprehensive evaluation while maintaining accuracy through the collective judgment of specialized evaluators.
Solution Approach 2:
The system enables self-service by automatically generating situation evaluations without requiring continuous ATM expert intervention for sample collection. The base evaluators autonomously assess situation factors using predefined criteria and historical data, reducing dependency on expert time while maintaining evaluation quality.
2Reliability
If traditional machine learning models require training samples and samples to be consistent in every aspect, then the model can be trained effectively, but samples from non-target sectors cannot be utilized which aggravates the lack of training samples
Solution Approach 1:
The patent creates a universal evaluation framework where base evaluators can process samples from both target and non-target sectors. The situation factor definitions and evaluation criteria are designed to be sector-agnostic, allowing the same evaluation mechanisms to function across different sectors and thus expanding the usable training sample pool.
Solution Approach 2:
The system transforms the evaluation approach by changing from sector-specific rigid matching to parameter-based flexible evaluation. By defining situation factors as standardized parameters with universal evaluation rules, the system can adaptively evaluate samples from different sectors without requiring exact sector matching, thereby utilizing diverse training data effectively.
3Productivity
If air traffic controllers handle large air transport volume, then the movement of goods and people accelerates, but the workload on air traffic controllers increases causing operational errors
Solution Approach 1:
The patent implements feedback mechanisms where the sector situation evaluation system continuously monitors traffic workload and provides real-time assessments to air traffic controllers. This feedback loop enables controllers to adjust their workload management based on objective situation evaluations, preventing overload conditions that lead to errors while maintaining high productivity.
Solution Approach 2:
The system replaces manual workload assessment with automated machine learning-based evaluation. The base evaluators objectively measure situation factors and generate workload assessments, substituting subjective human judgment with consistent automated analysis. This reduces the cognitive burden on controllers while improving the accuracy of workload monitoring.
Data Source
AI summary
A sector situation (SS) evaluation framework is based on knowledge transfer and is specifically applicable for small-training-sample environment. The SS evaluation framework is able to effectively mine knowledge hidden within the samples of both target and non-target sectors, and properly handle the integration between the knowledge derived from different sectors. This framework includes three main steps: (1) sufficiently mine the knowledge within the samples of the target sector using the strategies of multi-factor subset generation and multi-base evaluator construction, and build target base evaluators; (2) precisely learn the knowledge in the samples of the non-target sectors using similar strategies for the target sector, together with a sample transformation, and build non-target base evaluators; and (3) efficiently integrate the target and non-target base evaluators based on evaluation confidence analysis of those base evaluators.