面向广义柔性并行测试调度的嵌入式优化策略

Embedded Optimization Strategy for Generalized Flexible Parallel Test Scheduling

摘要:

针对广义柔性并行测试调度（GFPTS）问题中群优化算法早熟收敛、慢速收敛、寻优精度低这三类缺陷，提出一种基于特征评分与云模型的调度优化策略（FSCB）。FSCB通过问题导向型特征评分机制实现种群自适应分层；融合关键工序提取与差异化云模型，构建14种个体更新方式以提升种群多样性；同时具备无缝嵌入、多态输出等特点。在五种不同工程场景对应的自建用例上，选择优缺点鲜明的麻雀搜索算法、秃鹰搜索算法、河马优化算法作为FSCB验证载体，结果表明，FSCB策略成功率达80~90%，综合优化效能达17~44%，嵌入后载体算法早熟收敛概率平均降低27%，慢速收敛概率平均降低8%，寻优精度提升24~55%。FSCB策略为中等规模复杂组合调度问题提供了一种有效增强方案。

Abstract:

To address three types of defects—premature convergence, slow convergence and poor optimization accuracy—of swarm intelligence optimization algorithms in solving the Generalized Flexible Parallel Test Scheduling (GFPTS) problem, a Feature Scoring and Cloud Model-based scheduling optimization strategy (FSCB) is proposed. FSCB achieves adaptive population stratification through a problem-oriented feature scoring mechanism. By integrating critical operation extraction with differentiated cloud models, it constructs 14 individual update modes to enhance population diversity. The strategy also features seamless embedding and polymorphic output capabilities. On five self-built test cases corresponding to different engineering scenarios, three representative swarm intelligence algorithms with distinct advantages and disadvantages—Sparrow Search Algorithm (SSA), Bald Eagle Search (BES), and Hippopotamus Optimization (HO)—are selected as carriers for validating FSCB. Experimental results show that the success rate of the FSCB strategy ranges from 80% to 90%, and its comprehensive optimization efficacy reaches 17%–44%. After embedding FSCB, the probability of premature convergence of the carrier algorithms decreases by an average of 27%, the probability of slow convergence decreases by an average of 8%, and the solution accuracy improves by 24%–55%. The FSCB strategy provides an effective enhancement scheme for medium-scale complex combinatorial scheduling problems.