基于TransMP模型的Web系统剩余寿命预测方法

针对当前软件剩余使用寿命预测方法忽略了多性能指标间所蕴涵寿命信息的问题,提出一种融合多性能指标Transformer(TransMP)模型的Web系统剩余寿命预测方法。首先,搭建内存故障型Web系统加速老化实验平台,创建包含内存使用量、响应时间和吞吐率性能指标的数据集;其次,考虑不同性能指标蕴涵老化特征信息的差异性,构造由多编码器-解码器组成的TransMP模型,将性能指标数据分别输入内存指标编码器、响应时间编码器和吞吐率编码器提取老化特征信息,再引入特征融合层进行信息融合;最后,将融合信息输入由掩码注意力-多头注意力结构构成的解码器,预测得到系统状态达到老化阈值的剩余寿命。实验结果表明,该Web系统剩余寿命预测方法与最优的SALSTM方法相比,均方根误差分别降低了12.0%、17.3%和13.2%,平均绝对误差分别降低了13.3%、21.0%和10.4%,证明了该方法的有效性。

Construction of Antisense Transforming Growth Factorβ_1 Gene and Its Effect on the Proliferation by Expression in Osteosarcoma Cells

To construct the antisense transforming growth factorβ1 (TGFβ1) gene and investigate the effect of TGFβ1 autocrine loop blockage on the proliferation of osteosarcoma cells. TGFβ1 cDNA was cloned by RT-PCR from human osteosarcoma cells (MG-63) and inserted into pcDNA3 to construct an antisense expression vector, which was dubbed pcDNA3-TGFβ1(-). MTT was used to detect the proliferation of osteosarcoma cells transfected by antisense TGFβl gene. Our results showed that the proliferation of the transfected osteosarcoma cells was suppressed markedly. It is concluded that TGFβ1 autocrine loop blockage in osteosarcoma cells could inhibit cell proliferation, which might be helpful for gene therapy of osteosarcoma.