Using the Decomposition-Based Multi-Objective Evolutionary Algorithm with Adaptive Neighborhood Sizes and Dynamic...

Area of Science:

• Geophysics and Remote Sensing
• Computational Intelligence and Evolutionary Computation

Background:

• Traditional atmospheric duct retrieval methods using weather balloons or rocket soundings are expensive and inefficient.
• Advancements in technology allow for atmospheric duct retrieval using Global Navigation Satellite System (GNSS) phase delay and propagation loss data.
• GNSS receivers on the ground can function as automatic sensors for atmospheric duct monitoring.

Purpose of the Study:

• To propose a novel hybrid decomposition-based multi-objective evolutionary algorithm with adaptive neighborhood sizes (EN-MOEA/ACD-NS) for improved atmospheric duct retrieval.
• To dynamically constrain optimization objectives to balance population diversity and convergence.
• To validate the algorithm's performance on benchmark test instances and a practical atmospheric duct retrieval problem.

Main Methods:

• Development of the EN-MOEA/ACD-NS algorithm, integrating decomposition-based multi-objective evolutionary algorithms (MOEA/D) with adaptive neighborhood size adjustments.
• Dynamic imposition of constraints on optimization objectives to manage the trade-off between solution diversity and convergence speed.
• Performance evaluation using standard metrics such as hypervolume (HV), inverted generational distance (IGD), and average Hausdorff distance (∆2) on CEC 2009 test instances.

Main Results:

• The EN-MOEA/ACD-NS algorithm demonstrated comparable performance to existing methods like MOEA/ACD-NS in benchmark tests.
• The algorithm effectively balances population diversity and convergence, leading to high-quality optimal solutions.
• Successful application to the practical problem of jointly retrieving atmospheric ducts with GNSS signals confirmed its feasibility.

Conclusions:

• The proposed EN-MOEA/ACD-NS algorithm offers a more efficient and accurate method for retrieving atmospheric ducts compared to traditional techniques.
• The adaptive neighborhood size strategy and dynamic objective constraints are key to the algorithm's effectiveness.
• The study highlights the potential of advanced evolutionary algorithms in solving complex atmospheric remote sensing problems using GNSS data.