基于GA-SVR代理模型的谷物风筛清选系统优化设计

Optimization design of the grain air-and-screen cleaning system based on the GA-SVR surrogate model

谷物风筛清选系统是收获机械的核心功能部件,直接影响清选作业的损失率和含杂率。为提升其性能,降低清选损失率和含杂率,基于计算流体动力学和离散单元法耦合技术(CFD-DEM)构建清选系统气固两相流数值模型,采用正交试验方法设计L25(54)仿真试验,并使用遗传算法优化参数的支持向量回归算法(GA-SVR)建立清选系统损失率、含杂率代理模型,然后基于带精英策略的非支配排序遗传算法(NSGA-II)对代理模型进行多目标优化,获得风力因素、振动因素和结构因素的较优方案:入风口风速为9.5 m/s,振幅为32 mm,振频为6 Hz,筛孔大小为17 mm。最后,经整机田间试验验证,采取优化方案进行收获作业时,损失率降为0.75%,含杂率降为0.92%。

The grain air-and-screen cleaning system is a core functional unit of harvesters,which directly affects both the loss rate and the impurity rate of cleaning operation.In order to improve the performance of the cleaning system as well as reduce the loss rate and the impurity rate,the numerical model of gas-solid two-phase flow in the cleaning system is set up based on the computational fluid dynamics and with the aid of the discrete element method of coupling approach(CFD-DEM).Furthermore,the L25(54)simulation experiment is conducted by means of the orthogonal test,and the support vector regression algorithm with optimal parameters based on the genetic algorithm(GA-SVR)is adopted to work out the surrogate models of loss rate and impurity rate respectively.Then,the key parameters of the cleaning system are optimized by means of the NSGA-II multiobjective optimization algorithm on the basis of these surrogate models.The optimization results show that when the wind speed is 9.5 m/s,the amplitude is 32 mm,the vibration frequency is 6 Hz,and the size of the screen hole is 17 mm,the loss rate and the impurity rate of the cleaning system are desirable.Finally,it’s verified by the field test that when the optimization scheme is adopted for harvesting operation,the loss rate is reduced to 0.75%and the impurity rate is reduced to 0.92%.