In the field of meteorology, atmospheric duct has important implications for the transmission of electromagnetic wave. When the electromagnetic wave signal is received by the signal receiving antenna of the global navigation satellite system (GNSS), the propagation loss and phase delay of the electromagnetic wave in the actual propagation process can be recorded, and the predicted values of the atmospheric dust parameters can be obtained through the inversion process. Atmospheric duct inversion problem can be modeled as a multi-objective optimization problem. Based on the classic MOEA/D algorithm, this paper designs an evolutionary multi-objective optimization algorithm for a single GNSS received signal, which introduces multiple population strategy and dynamic neighborhood mechanism. This paper also compares and analyzes the proposed algorithm with the classical evolutionary optimization algorithms through experiments. The experimental results show that the algorithm has higher accuracy and can better solve the atmospheric duct inversion problem.