Deep Unsupervised Airspace Complexity Evaluation via Stacked Autoencoder
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current airspace complexity evaluation methods rely heavily on manual calculations and labeled data, leading to inaccurate results, high costs, and difficulty in real-time evaluation, as they fail to effectively classify airspace complexity due to high dimensionality and non-linear relationships in data.
Innovation Solution
A deep unsupervised learning approach using a stacked autoencoder and t-distribution-based loss function to evaluate airspace complexity without labeled data, enabling real-time classification through nonlinear dimension reduction and clustering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional supervised learning methods are used for airspace complexity evaluation, then classification accuracy can be improved with sufficient labeled data, but the cost of obtaining massive labeled samples increases significantly in terms of time and labor
Solution Approach 1:
The patent employs unsupervised learning algorithms that enable the system to automatically learn from unlabeled airspace complexity data without requiring manual annotation. The model self-organizes to identify patterns and classify complexity levels, eliminating the need for expensive and time-consuming labeled data preparation while maintaining effective evaluation capabilities
2Productivity
If traditional unsupervised learning methods like PCA are used for dimension reduction, then computational efficiency is improved, but the ability to capture non-linear relationships between airspace complexity factors is lost
Solution Approach 1:
The patent transforms the approach by using deep neural network-based unsupervised learning methods that can model non-linear relationships in the data. By changing from linear dimension reduction techniques to non-linear deep learning architectures, the system maintains computational efficiency while significantly improving its ability to capture complex non-linear patterns in airspace complexity factors
Solution Approach 2:
The patent combines multiple unsupervised learning techniques and architectural components to create a composite evaluation system. This includes integrating autoencoders for feature extraction, clustering algorithms for classification, and attention mechanisms for feature importance weighting, creating a robust system that handles non-linear relationships effectively
3Adaptability or versatility
If manual calculation methods based on controller experience are used, then adaptability to various airspace situations is improved, but the evaluation process becomes time-consuming and cannot achieve real-time assessment
Solution Approach 1:
The patent replaces the manual mechanical calculation process with an automated computational system based on unsupervised learning. The system automatically processes airspace data, applies learned patterns from training, and generates complexity evaluations in real-time, substituting human manual operations with algorithmic processing that maintains adaptability while achieving real-time performance
Data Source
AI summary
This application provides an airspace complexity evaluation method based on deep unsupervised learning for air traffic management, which includes the following parts. A stacked autoencoder is used to establish an airspace complexity evaluation model. Input the airspace complexity factors into the stacked autoencoder to obtain the low-dimensional embedded representations of the airspace complexity factors. Cluster the low-dimensional embedded points to capture the centroids of the airspace complexity data. The application utilizes the soft assignment distribution and real assignment distribution of the embedded representations to construct a training loss function which optimizes the airspace complexity evaluation model by gradient descent algorithm. The trained airspace complexity evaluation model and the three obtained cluster centroids describing the airspace complexity level are used to evaluate the current airspace complexity.