论文部分内容阅读
针对结构有限元分析中模型精度低的问题,提出了响应面优化的模型修正方法.利用优化拉丁方试验设计样本点,F值检验筛选出修正参数构造不完全4阶多项式响应面并检验响应面有效性.引入遗传算法以响应面计算值与试验值的残差为目标来优化修正参数,将修正值代入初始模型中得到修正模型.三自由度系统数值算例,验证了模型中参数准确、参数不存在和参数有误差3种工况下,响应面优化方法的模型修正能力.以GARTEUR飞机模型为对象,筛选显著性参数进行模型修正,结果表明修正后模型不但在修正频段和预测频段具有良好的复现和预测能力,还能预测结构局部修改后的频率,证实了经过参数筛选后响应面优化的模型修正有效性.
Aiming at the problem of low accuracy of model in structural finite element analysis, a method of model optimization for response surface optimization is proposed. By using the optimized Latin square test design sample points, F value test is used to filter out the modified parameters to construct an incomplete 4th order polynomial response surface and verify the response surface Effectiveness.The genetic algorithm was introduced to optimize the correction parameters and calculate the residuals of the calculated values and experimental values of the response surface.The modified three-degree-of-freedom system was used to verify the accuracy of the model, Parameters are not exist and the parameters are error under the three kinds of working conditions, response surface optimization method of model correction capability.GARTEUR aircraft model as the object, screening significant parameters for model correction, the results show that the modified model not only in the correction band and the prediction band has Good reproducibility and predictive ability, and predictions of locally modified frequencies of structures, confirming the validity of response surface-optimized models after parameter screening.