研究多领域仿真优化中SQP算法的并行处理与调度策略,提出了基于多领域仿真的SQP并行优化问题中的抽象调度模型即等式约束离散变量优化模型,对算法理论的可行性做了深入探讨;采用机群系统构建了并行仿真优化环境,在自主研发的多领域统一建模与仿真平台MWorks下实现了并行优化模块。以F14战机简易模型的控制参数优化为例,验证了该方法的有效性。
A parallel processing and scheduling strategy of SQP algorithm in multi-domain simulation optimization was studied. We proposed the abstract scheduling model in parallel optimization problem: an discrete variable optimization model with equality constraints. Details of cluster of workstations were given, a thorough theoretical feasibility of parallel algorithm. Under the uniform modeling and simulation platform for multi-domain systems, MWorks, parallel optimization and process were implemented. The efficiency of SQP optimization of parallel technique was proved by an example of the design optimization of a simple F14 aircraft model.