基于粗粒度并行遗传算法的隔震层参数优化

Parameter optimization of isolation layer based on coarse-grained parallel genetic algorithm

摘要:

针对基于经典遗传算法的隔震层参数优化方法效率不高的问题,提出一种基于粗粒度并行遗传算法的隔震层参数优化方法。利用Python的多进程机制和Python与ETABS的交互,实现CPU各核同时调用ETABS并进行遗传操作,最后通过一个隔震工程的实例进行验证。结果表明:采用粗粒度并行遗传算法进行隔震层参数优化,与原设计结果相比,优化后的隔震结构性能更优;同时,用10核CPU计算,与经典遗传算法相比,该方法既能准确得出全局最优解,又可显著提高优化效率,加速比约为6,可基本满足隔震工程设计的及时性需求,具有较好的工程应用价值。

Abstract:

Given the low efficiency of the parameter optimization of the isolation layer based on a classical genetic algorithm, a parameter optimization method for the isolation layer based on a coarse-grained parallel genetic algorithm was proposed in this paper. Using the multiprocess mechanism of Python and the interaction between Python and ETABS, each core of the CPU can simultaneously perform ETABS to implement dynamic time-history analysis of isolation and nonisolation structures and use the CPU multiprocess to realize parallel genetic operation in genetic algorithms. According to the test results, compared with the original design results, the isolation structure optimized by a coarse-grained parallel genetic algorithm has a better performance. Meanwhile, compared with the classical genetic algorithm, the proposed method can not only accurately obtain globally optimal solutions but also substantially improve the optimization efficiency. The speedup ratio is approximately 6, which can meet the prompt demand for design in isolation engineering. Therefore, this method has a notable application potential.