改进元胞多目标遗传算法在机床主轴优化中的应用

摘要

针对目前采用多目标元胞遗传算法解决具有整数决策变量和实数决策变量问题时，容易出现因参数设置的盲目性而导致求解失效或效果较差的问题，以机床主轴优化设计问题为研究对象，建立了以总体积最小和挠度最小为设计目标的多目标优化设计数学模型。在采用多目标元胞遗传算法( MOCell)对问题模型进行求解时，结合问题特点对个体编码和变异算子进行了重新设计，同时阐明了通过正交试验优化遗传参数配置的思想。通过实验仿真，所得Pareto前端分布均匀、覆盖范围较为宽广，说明修改后的算法求解此类问题行之有效。%When solving the problems that have integer decision variables and float decision variables with multi-objective cellular genetic algorithm, it is sometimes failed or gets poor results due to bad parameter settings. Taking the machine spindle optimization design as the study object, a multi-objective optimization design model is established with the objectives to get minimal volume and minimal deflection. When the multi-objective cellular genetic algorithm( MOCell) is applied, the individual code and the crossover opera-tor are remodified considering the characteristics of the solved problem, and at the same time, the idea of the orthogonal experiment is introduced into the algorithm. Through the simulation, a Pareto front is obtained, which distributes evenly and spreads widely. The result indicates that the modified algorithm works well with this kind of problems.