The improved microstructure and enhanced elevated temperature mechanical properties of Ti-44Al-5Nb-(Mo,V,B)alloys were obtained by vacuum arc re-melting(VAR)and primary annealing heat treatment(HT)of 1260℃/6 h/Furnac...The improved microstructure and enhanced elevated temperature mechanical properties of Ti-44Al-5Nb-(Mo,V,B)alloys were obtained by vacuum arc re-melting(VAR)and primary annealing heat treatment(HT)of 1260℃/6 h/Furnace cooling(FC).The phase transformation,microstructure evolution and tensile properties for as-cast and HTed alloys were investigated.Results indicate that three main phase transformation points are determined,T_(eut)=1164.3℃,T_(γsolv)=1268.3℃and T_(βtrans)=1382.8℃.There are coarse lamellar colonies(300μm in length)and neighbor reticular B2 andγgrain(3-5μm)in as-cast alloy,while lamellar colonies are markedly refined and multi-oriented(20-50μm)as well as the volume fraction and grain sizes of equiaxedγand B2 phases(about 15μm)significantly increase in as-HTed alloy.Phase transformations involvingα+γ→α+γ+β/B2 and discontinuousγcoarsening contribute to the above characteristics.Borides(1-3μm)act as nucleation sites forβ_(eutectic) and produce massiveβgrains with different orientations,thus effectively refining the lamellar colonies and forming homogeneous multi-phase microstructure.Tensile curves show both the alloys exhibit suitable performance at 800℃.As-cast alloy shows a higher ultimate tensile stress of 647 MPa,while a better total elongation of more than 41%is obtained for as-HTed alloy.The mechanical properties improvement is mainly attributed to fine,multi-oriented lamellar colonies,coordinated deformation of homogeneous multi-phase microstructure and borides within lamellar interface preventing crack propagation.展开更多
流标签是当前多标签学习领域中一个较新颖的挑战性问题,存在标签空间未定、标签数量不断增加甚至趋于无穷等问题.在多标签学习的特征选择中,每当有新的标签到达时标签空间都将发生改变,传统的多标签特征选择算法需重新进行特征选择,所...流标签是当前多标签学习领域中一个较新颖的挑战性问题,存在标签空间未定、标签数量不断增加甚至趋于无穷等问题.在多标签学习的特征选择中,每当有新的标签到达时标签空间都将发生改变,传统的多标签特征选择算法需重新进行特征选择,所以不适用.针对此问题,采用将流标签进行分组批量处理的方式,并考虑标签之间的相关性,提出一种新的流式多标签特征选择方法,考虑分组后每组标签内部潜在的关联结构和不同标签组之间的标签差异性,赋予每组标签不同的权重来计算每个特征与标签空间的模糊互信息.同时,结合mRMR(Max-Relevance and Min-Redundancy)的特征选择策略进行冗余特征的剔除,从而挑选最优的特征子集.该方法同时适用于固定标签空间和流式标签空间中的特征选择问题.最后,选取八个多标签基准数据集,采用四种评价指标与已有相关的多标签特征选择方法进行对比实验,实验结果证明了提出方法的有效性和高效性.展开更多
基金Funded by the National Natural Science Foundation of China(No.52071065)Fundamental Research Funds for the Central Universities(No.N2007007)。
文摘The improved microstructure and enhanced elevated temperature mechanical properties of Ti-44Al-5Nb-(Mo,V,B)alloys were obtained by vacuum arc re-melting(VAR)and primary annealing heat treatment(HT)of 1260℃/6 h/Furnace cooling(FC).The phase transformation,microstructure evolution and tensile properties for as-cast and HTed alloys were investigated.Results indicate that three main phase transformation points are determined,T_(eut)=1164.3℃,T_(γsolv)=1268.3℃and T_(βtrans)=1382.8℃.There are coarse lamellar colonies(300μm in length)and neighbor reticular B2 andγgrain(3-5μm)in as-cast alloy,while lamellar colonies are markedly refined and multi-oriented(20-50μm)as well as the volume fraction and grain sizes of equiaxedγand B2 phases(about 15μm)significantly increase in as-HTed alloy.Phase transformations involvingα+γ→α+γ+β/B2 and discontinuousγcoarsening contribute to the above characteristics.Borides(1-3μm)act as nucleation sites forβ_(eutectic) and produce massiveβgrains with different orientations,thus effectively refining the lamellar colonies and forming homogeneous multi-phase microstructure.Tensile curves show both the alloys exhibit suitable performance at 800℃.As-cast alloy shows a higher ultimate tensile stress of 647 MPa,while a better total elongation of more than 41%is obtained for as-HTed alloy.The mechanical properties improvement is mainly attributed to fine,multi-oriented lamellar colonies,coordinated deformation of homogeneous multi-phase microstructure and borides within lamellar interface preventing crack propagation.
文摘流标签是当前多标签学习领域中一个较新颖的挑战性问题,存在标签空间未定、标签数量不断增加甚至趋于无穷等问题.在多标签学习的特征选择中,每当有新的标签到达时标签空间都将发生改变,传统的多标签特征选择算法需重新进行特征选择,所以不适用.针对此问题,采用将流标签进行分组批量处理的方式,并考虑标签之间的相关性,提出一种新的流式多标签特征选择方法,考虑分组后每组标签内部潜在的关联结构和不同标签组之间的标签差异性,赋予每组标签不同的权重来计算每个特征与标签空间的模糊互信息.同时,结合mRMR(Max-Relevance and Min-Redundancy)的特征选择策略进行冗余特征的剔除,从而挑选最优的特征子集.该方法同时适用于固定标签空间和流式标签空间中的特征选择问题.最后,选取八个多标签基准数据集,采用四种评价指标与已有相关的多标签特征选择方法进行对比实验,实验结果证明了提出方法的有效性和高效性.