In order to optimize the crashworthy characteristic of energy-absorbing structures, the surrogate models of specific energy absorption （SEA） and ratio of SEA to initial peak force （REAF） with respect to the design...In order to optimize the crashworthy characteristic of energy-absorbing structures, the surrogate models of specific energy absorption （SEA） and ratio of SEA to initial peak force （REAF） with respect to the design parameters were respectively constructed based on surrogate model optimization methods （polynomial response surface method （PRSM） and Kriging method （KM））. Firstly, the sample data were prepared through the design of experiment （DOE）. Then, the test data models were set up based on the theory of surrogate model, and the data samples were trained to obtain the response relationship between the SEA ＆amp; REAF and design parameters. At last, the structure optimal parameters were obtained by visual analysis and genetic algorithm （GA）. The results indicate that the KM, where the local interpolation method is used in Gauss correlation function, has the highest fitting accuracy and the structure optimal parameters are obtained as： the SEA of 29.8558 kJ/kg （corresponding toa=70 mm andt= 3.5 mm） and REAF of 0.2896 （corresponding toa=70 mm andt=1.9615 mm）. The basis function of the quartic PRSM with higher order than that of the quadratic PRSM, and the mutual influence of the design variables are considered, so the fitting accuracy of the quartic PRSM is higher than that of the quadratic PRSM.展开更多
基于变量预测模型的模式识别(Variable predictive model based class discriminate,简称VPMCD)方法在训练过程中是用多项式响应面(Polynomial Response Surface,简称PRS)法来建立预测模型的,然而PRS法的模型拟合精度不能随训练样本容...基于变量预测模型的模式识别(Variable predictive model based class discriminate,简称VPMCD)方法在训练过程中是用多项式响应面(Polynomial Response Surface,简称PRS)法来建立预测模型的,然而PRS法的模型拟合精度不能随训练样本容量的增加而显著提高。针对这一缺陷,将原方法中的PRS方法进行了改进,提出了基于改进多项式响应面(Improved Polynomial Response Surface,简称IPRS)的VPMCD方法,并将其应用于滚动轴承故障诊断。通过实验,将原方法和改进方法在训练样本容量不同情况下的模式分类精度进行对比,结果表明,相对于原VPMCD方法,改进的VPMCD方法不仅具有更好的模式分类效果,而且其分类精度随训练样本容量的增加提高得更明显。展开更多
基金Project(U1334208)supported by the National Natural Science Foundation of ChinaProject(2013GK2001)supported by the Fund of Hunan Provincial Science and Technology Department,China
文摘In order to optimize the crashworthy characteristic of energy-absorbing structures, the surrogate models of specific energy absorption （SEA） and ratio of SEA to initial peak force （REAF） with respect to the design parameters were respectively constructed based on surrogate model optimization methods （polynomial response surface method （PRSM） and Kriging method （KM））. Firstly, the sample data were prepared through the design of experiment （DOE）. Then, the test data models were set up based on the theory of surrogate model, and the data samples were trained to obtain the response relationship between the SEA ＆amp; REAF and design parameters. At last, the structure optimal parameters were obtained by visual analysis and genetic algorithm （GA）. The results indicate that the KM, where the local interpolation method is used in Gauss correlation function, has the highest fitting accuracy and the structure optimal parameters are obtained as： the SEA of 29.8558 kJ/kg （corresponding toa=70 mm andt= 3.5 mm） and REAF of 0.2896 （corresponding toa=70 mm andt=1.9615 mm）. The basis function of the quartic PRSM with higher order than that of the quadratic PRSM, and the mutual influence of the design variables are considered, so the fitting accuracy of the quartic PRSM is higher than that of the quadratic PRSM.
文摘基于变量预测模型的模式识别(Variable predictive model based class discriminate,简称VPMCD)方法在训练过程中是用多项式响应面(Polynomial Response Surface,简称PRS)法来建立预测模型的,然而PRS法的模型拟合精度不能随训练样本容量的增加而显著提高。针对这一缺陷,将原方法中的PRS方法进行了改进,提出了基于改进多项式响应面(Improved Polynomial Response Surface,简称IPRS)的VPMCD方法,并将其应用于滚动轴承故障诊断。通过实验,将原方法和改进方法在训练样本容量不同情况下的模式分类精度进行对比,结果表明,相对于原VPMCD方法,改进的VPMCD方法不仅具有更好的模式分类效果,而且其分类精度随训练样本容量的增加提高得更明显。
