Multistrategy Improved Aquila Optimizer for Test Case Prioritization

Traditional heuristic algorithms often fall into local optima and converge slowly when test case prioritization is addressed in regression testing, making them inadequate for complex real-world scenarios. The Aquila optimizer, a novel metaheuristic algorithm, demonstrates strong global exploration capability but still faces limitations, including insufficient exploitation capability and slow convergence. To overcome these challenges, a multi-strategy improved chaotic Cauchy inverse cumulative distribution Aquila optimizer for test case prioritization is proposed. First, a logistic–sine–cosine composite chaotic mapping is introduced during the initialization phase of the Aquila optimizer to increase population diversity. Second, the mutated random walk strategy is used to improve global exploration, further enhancing the global search ability of the Aquila optimizer. Moreover, during the narrowed exploration and narrowed exploitation phases, the Cauchy inverse cumulative distribution flight replaces the Lévy flight strategy to reallocate individual positions, strengthening individuals’ optimization capability and preventing the algorithm from becoming trapped in local optima. Finally, in the later iteration stage, the specular reflection learning strategy is used to perturb the optimal individual positions and improve the Aquila optimizer’s convergence accuracy and comprehensive optimization performance. Five Java projects were selected from the Defects4J benchmark datasets to conduct comparative experiments with the Aquila optimizer and seven other metaheuristic algorithms. The results demonstrate the effectiveness and superiority of the improved algorithm in test case prioritization. It achieves average improvements of approximately 4.96% in the average percentage of fault detection, 3.82% in the average percentage of block coverage, and 5.64% in the average percentage of decision coverage, enabling faster coverage of code blocks and branches. The results provide an efficient priority sorting solution for complex regression testing scenarios.