具有反向学习的多目标元胞遗传算法的空间桁架结构优化设计

Optimization Design of Space Truss Structure Based on Multi-objective Cellular Genetic Algorithm with Opposition-based Learning

为了解决具有多约束的桁架结构问题,提出一种具有反向学习的多目标元胞遗传算法应用于空间桁架结构多目标优化设计中。根据分析元胞遗传算法特点,引入一种反向学习策略、差分进化策略和约束处理技术。通过标准测试函数对比分析,算法能很好地保持Pareto解集的收敛性和均匀性。针对空间桁架结构优化的数学模型,采用实数编码和个体修正方法,将该算法对72杆空间桁架优化问题进行求解,并与MOCell的优化结果进行比较。结果表明,新算法获得的Pareto解集更加均匀,极端点值域更宽广,具有一定的工程实用性。

In order to solve the problems of truss structures with multi-constraint,a multi-objective cellular genetic algorithm based on opposition-based learning（ DECell-OBL） was proposed for solving the space truss structure multi-objective optimization design.Based on the analysis of the cellular characteristics of genetic algorithm,the opposition-based learning strategy,differential evolution strategy and constraint handling technology were introduced into the algorithm.Through the comparison and analysis of other algorithms by the benchmarks,the new algorithm could keep the convergence and uniformity of the Pareto Set.According to the mathematical model of spatial truss structure optimization,DECell-OBL was used real encoding and individual correction method to solve the problem of 72 bar truss space,and compared with MOCell.The simulation results showed that the new algorithm to obtain the Pareto solution set is more uniform and extreme point range wider.DECell-OBL has more certain engineering practicability.