Identification of region of interest in test sample space based on adaptive multi-strategy particle swarm optimization

• Four characterization metrics are proposed for the multidimensional characterization of the region of interest(ROI) in the test sample space. • An algorithm is proposed for identifying the ROI under specific constraints, which can accurately detect the ROI in the test sample space. • A method for validating the ROI is introduced, which is validated through various dimensional arithmetic cases, demonstrating better recognition of the ROI and its application to radar anti-jamming scenarios. Aiming at the problem of identifying the Region of Interest(ROI) in a high-dimensional test sample space with complex constraints, this paper proposes an adaptive multi-strategy particle swarm optimization (MS-PSO) method for ROI identification. First, a new comprehensive characterization index is developed, based on the Pareto dominance relation, to quantitatively describe the critical regions that encompass the four key characteristics of the test sample space: ruggedness, regional variability, global similarity, and global complexity. An adaptive MS-PSO is then designed, incorporating strategies such as adaptive parameters, restart mechanisms, and clustering to enhance the search diversity. This approach effectively addresses the issue of identifying the critical regions within the test sample space under complex constraints. The methodology is validated through numerical examples and radar anti-jamming performance test cases. The results demonstrate that, compared to three existing optimization algorithms, the proposed method improves key performance metrics such as the mean squared error of model construction, the coefficient of determination, and other region characteristics, thereby providing support for the formulation of design strategies in subsequent test designs.